LIBRA Jr.) to -1- I
Or LIBRA Jr.) to -1- I
APOSTATIC SELECTION: THE RESPONSES OF
WILD PASSERINES TO ARTIFICIAL POLYMORPHIC PREY
by
John Ashdown Allen
Thesis submitted for the Degree of Doctor of
Philosophy University of Edinburgh, December 1972 -2-
ABSTRACT
The iaintonanco of non-mimetic colour polymorphism has yet to be satisfactorily explained. Apotatic selection is one possible answer: sight-dependent predators may tend to form 'searching images' for common morphs and concentrate on those varieties in preference to rarer ones • (Chapter t).
Some evidence for apostatic selection exists, but it is mostly indirect and incomplete'. (Chapter II).
The hypothesis was tested by experiments with wild passerine birds in their normal surroundings. The prey wore lard-and-flour
'baits' which were usually green or brown and cylindrical (0.7cm long x 0.7cm diameter). Data were obtained by direct or indirect observation of 'populations' containing the colours in known proportions. Eaton baits were replaced frequently to ensure that the ratios wore kept constant. Five cots of experiments wore carried out. (chapter II!).
In one set (Chapter IV) baits were at maximum density. Mon unequal numbers of greens and browns were presented there was a significant overall tendency for the ftrer forms to be removed.
Individual blackbirds Tur'dua morula tended to concentrate on one colour alone within visits. Presentation of greens and browns in equal numbers gave no evidence that the colours differed in taste.
Further experiments with small green and brown baits and small rod and yellow baits confirmed that rare forms are preferred at maximum density. -3-
In the remaining sets of oxperimonte baits tiore at a density of tio per square metre. Populations of randomly distributed groans and browns were presented on grass or soil backgroncIe.
A second sot of axperiments (Chapter V) revealed that birds can bocoize conditioned to searching for either green or broin baits. Mid birds were fed with one colour alone and toro then given populations with both colours in equal proportions. In all eight operiments the familiar colour was strongly preferred. In three cases the preferences woro reversed by a further training period on the'unfamiliar' colour.
A third cot (Chapter VI) shoed that psscorinea on average tend to take more of the common variety than aEpected on the bcsi of constant selection • In each experiment populations tore presented either with greens nine times as comian as brome folloed by populations with brome nine times as common as greens or in the rerso order. The initial study revealed heterogeneity within the data and repeat experiments wora therofere carried Cut. The data frW the fourteen experiments wam consistent • In thirteen cases the common colour was preferred ovoll, despite different environmental conditions.
A fourth sot investigated the behaviour of individual blackbirds and songtbr'uahos Turdua philociclos. One experiment
(Chapter VII) showed that the three birds involved preferred brotmo • In two cases these preferences were very strong and the birds otartod to oat greens in quantity only tihen brww mere very rare • This behaviour woo preceded by an increase in the tendency to 'imprint-pew. The three birds appeared to differ markedly -'4 -
in their rasponsas. Too further eporirents (Chapter VIII) with a majority of 'nc& birds confirmed the above findings and also provided evidence of searching images. At the and
of the third experiment populations of green, browns and khstde were offered. The khakis were initially overlooked.
In a fifth set (Chapter IX) greens, browns and various shades of khaki were used. Two series of experiments tooted the
specificity of conditioning. Birds were familiarized with one
colour and were than offered populations with the 'familiar'
11 bait and a visually similar colour in equal proportions. They were then trained on the 'unfamiliar' colour and given a choice as before. The results showed that birds can become conditioned
to a particular colour; this conditioning can be very specific, but reversible • In two further experiments an attempt was made to see whether predators Can promote polyorphietm. Blackbirds and songthrusbes wore offered populations containing nine morphs ranging from green to brown and in proportions that fitted the normal distribution. These praportionQ were altered regularly to accord with the freuoncioa that z'oiaatnod after a fixed percentage of baits had been eaton • One eporiment showed that the browner baits were preferred. The second, experiment was preceded by a period of familiaritation with the cotxnzonoet khaki morph and showed an increase in the variance of the population.
The distribution became bimodal • This disruptive effect was probably a result of the training, natural brown preferences, and the heterogeneity of behaviour of the birds.. Searching images
can therefore be very specific and under certain conditions predators may be capable of causing polymorphism in prey species. -5--
The oxporixita1 dasign is conidosd to resemblo a natural situation. It is ocnàludod that the results support the hyothoio of apoetatic Ooli3ction • The findingo have Implications fbr the conditions under which spostatic selection is likely to act and point the way to futura rosGarch i It is ougostod that a no of different typea of opoetcxtie polymorphism odot. (Chapter X). -6 -
PREFACE
My aim in this work has been to test a hypothesis. It has been suggested that variability within a population of a palatable prey species can be maintained by 'epostatic selection' by predators, the meaning of which will become clear later. This posibility has been investigated by presenting populations of artificial prey to wild passer'ino birds in their normal surroundings. The exorcise is therefore oituatod in the area of zoology where animal behaviour and population genetics overlap. We are concerned with the behaviour of predators and more important, with its effect an variability in the prey population.
The thesis is organised into ten Chziptorc. The first is introductory and provides the background to the problem. Existing evidence for apostatic selection is reviewed in the second Chapter.
A description of the general experimental approach is given in the third and the experiments themselves are discussed in the following seven Chapters. Bulky data are tabulated in appendices at the ends of the relevant Chapters. The final Chapter focuses on the main conclusions and points out possible lines of future research.
Acknowledgements
First and foremost I thank my supervisor Professor Bryan Clarke for his constant encouragement, criticism and discussion throughout all stages of the work.
I am also grateful to many other friends and colleagues who have discussed aspects of the work with me. In particular I must mention those members of the 'population genetics group' who were - 7 - at-the Department of Zoology, University of Edinburgh, between
1967 and 1971. I am indebted to Professor J.M. Mitchison and
Professor P.M.B. Walker for the facilities that I have enjoyed in their Department. The research was supported by a Research
Studentship from the Science Research Coulicil, to whom I extend my appreciation. Various other people gave help and I thank them all. They include those who allowed me to work on their land: Professor
Bryan and Mrs. Ann Clarke; the late Mr. J. Ewing; my parents, Professor and Mrs. P. Allen; and the farmers of Midlothian who allowed ma access to their fields. I am especially grateful to Miss Nan Brownlie, Mr. Kim Howell, Mrs. A. Almeida and Mrs. A. Patel who typed the first draft, and in particular Mrs. D. M.Powell who typed the thesis in its present form. The final typing, assembling and binding was done in Reading while I was in Dar es Salaam and I am very much indebted to Mrs. Powell for handling the arrangements at that end. My wife, Eleanor, gave much appreciated assistance in checking the final typescript. Mr.
Bob Brown photographed most of the Figures and Mr. David Briscoe
photographed the Colour Plates. Finally, I thank a motley crowd of people without whose
spiritual support this thesis would never have been finished. These
include Eleanor, Boris and Chris, John Clarke, Bones, Cohn Craig,
David and Jane, Captain Beefheart, and latterly Michael, John and Jan in Dar. All these people kept my sanity at times when prospects
seemed bleak.
-8-
CONTENTS Page
ABSTRACT 0 ,0 . 0 . • oo.o.... 000 000 2
PREFACE ..... 00006 ...... 0 ...... 6 Acknowledgements ...... 6 TABLESTEXT -FIGURES, PLATES ...... , ...... 13
TABLES. 0 0 0 0 00 0 0 ...... 00 ...... 0 0 0 00 0 4 13
TEXT-FIGURES ...... 0 0 0 • , • • • . . . • • • ...... 15
3 • PLATES ...... • • • • • • 0 0 0 0 0 0 0 0 4 ...... 16
CHAPTER I GENERAL INTRODUCTION ...... 17
1. COLOUR POLYMORPHISM AND ITS MAINTENANCE ...... o.... 18
2. SELECTION BY PREDATORS , ...... • 0 • • ...... 21
General 0 • • • • , • 0 ...... • , • 0 0 0 0 0 0 0 0 0 0 0 0 • 0 0 00 0 • 0 0 21
Apostatic selection • • • • • • • •• • 00 0 0 0 0 24 3. APOSTATIC SELECTION AND SEARCHING IMAGES 26
Searching image 0 0 00 • • 0 • • • ...... 0 • • • • ,0 26
Searching images and colour polymorphism 0 00 00 28
CHAPTER II THE EVIDENCE FOR APOSTATIC SELECTION ,.o....,, 29
1 . IN'rIDtJc'rIoN 0040000 • ...... , • 0 • • 30 2 . MIXED COLONIES OF CEPAEA ...... 31
3. PREDATOR-PREY STUDIES ...... 0 ...... • • 32 (a) Introduciñg the evidence ...... 32 (b) Do predators become conditioned to searching for specific palatable prey? ...... 33
( t) General evidence ...... 000 33 (ii) Natural polymorphic prey •...... 38 (c) Do predators become conditioned to searching for the commonest types in a mixture of palatable prey ...... 141 General evidence ...., ...... , ...... ,, 41 Natural polymorphic prey ...... , 46
4 . CONCLUSIONS • • • • • • 0 0 0 0 0 0 • • • ...... '47
(a) Conditioning to a specific prey •...... 000 47 Specificity of conditioning ., 47
Apostatic selection 0 ..... 00 000 0000 ...... 48
CHAPTER III GENERAL MATERIALS AND METHODS ...... 00 ...... '49
INTRODUCTION ...... 000 • •. • ...... •. .. . 50 CHOICE OF EXPERIMENTAL SYSTEM ...... ,....., 50 BIRDS •.,...,...... 00000*0000 ...... , ...... 51 4 , PREY ...... 0 ...... 53 Morphology ...... 0 0 .....0 0 0 00 ...... 0 0 • • 53 Manufacture ...... 35 ENVIRONMENTS ...... 0 0 ...... 55 TECHNIQUES ...... 0 0 0 0 ...... 0 0 0 0 • 0 0 0 0 ...... 56 (a) Presentation ofprey ...... 56 (1) Experimental plot ...... 56 (ii) Randomization of prey distribution 58
-9-
Page
(b) Recording of data ...... 60 7. PLAN OF EXPERIMENTS ...... , . . 61
CHAFFER IV MAXIMUM-DENSITY EXPERIMENTS ...... •... 64 1.. INTRODUCTION a . . , a • . . . a . , . , . , . ...... a o a a a a a a • 65 2. MATERIALS AND METHODS ...•...... 67 Series 1 ...... 67 Series 2 , ...... 67 3 . RESULTS . , • * • , , , • ...... 70 Series 1...... ,., 70 ( i) Synopsis ...... ,..., 70 (ii) Behaviour of birds .., ...... , 72 Series 2 . . . . . * ...... 74 4 . DISCUSSION ...... . • ...... 78 Colour-preferences and searching image ., 78 Së1ectionof rare prey ...... 80 5 . SUMMARY ...... 82
CHAPTER V CONDITIONING TO GREES OR BROWNS ,,...... , 84 1. INTRODUCTION . • ...... • . . • 85 2.PRELThINARY EXPERIMENT ...... 1 ...... 0 85 Procedures...... 85 Results and discussion .,...... , 87 S . REPEAT EXPERIMENTS . . . . . -...... 87 Plan ...... , . a 87 Materials and methods ...... 88 (1) Locations . . . 4 • , • , • • • • • • ...... 88 (ii) Procedures •.*., ...... 88 (ill) Predators ...... ,.a, 92 Results ...... • * •, • *• ...... a...... 92 4. DISCUSSION ...... * ...... 96 Comparison between preliminary and repeat experiments ...... 96 Effect of natural preferences 96 Conditioning .. . . , ...... , . . . . , ...... 97 5 . SUMMARY ...... ,...... •, ...... 100
CHAPTER VI 9:1 EXPERIMENTS: PREDATION BY GROUPS OF BIRDS...... * .. * . . . 101 INTRODUCTION ...... , . * . . . * ...... * ...... 102 PRELIMINARY EXPERIMENTS • ...... 102 Experiment 1: presentation of 9G:1B populations followed by 1G:9B populations. 102 (1) Materials and methods...... ,. 102 (ii) Results and-discussion ...,...,..... 104 Experiment 2.1: presentation of 1G:9B populations followed by 90:1B populations. 108 Materials and methods ...... 108 Results and discussion ...... ,,., 108
- 10 -
Page (c) Experiment 4: presentation of 1G: 9B populations followed by 1G:1B and 9G:1B populations ...... 110 (1) Materials and methods ...... ,..., 110 (ii) Results and discussion ...... , 110
3 . REPEAT EXPERIMENTS • • • 0 0 0 0 0 0 0 • 0 00 • 0 0 0 00 0.0 111
(a) Plan of work • • , . , . , ...... • * 0 * • • • , • 112 (1 ) Series 3 . . . • • • , • , . . * . • • • • • • • • 112
Series 5 . . • • • 0 • • 0 0 0 •, ...... 1.12 Series 6 ...... 113 (b) Experiments and results •...... 113
4. SUMMARY OF DATA ...... 0 0 0 ...... 113 S. DISCUSSION , ...... , . 116 Apostatic selection ...... , 116 Colour preferences ...... 118
Conditioning •,.0000...... , 121
6. SUMMARY ...,,...,....,,...... 00000000000 122
APPENDIX A: TABLES 22-32 0000 ...... ,...,...., 123 (a) Key to Tables ,, ...... • ...... 123 (b) Tables and expsrintsnts 123 (1) Presentation of 9G:1B populations followed by lG:9B populations 125 (ii) Presentation of 1G:9B populations followed by 9G:LB populations .,. 137
CHAPTER VII THE RESPONSES OF INDIVIDUAL TURDIDAE:
PART 1 0 a , ...... 0 • 0 o . . . . 146
INTRODUCTION 0 • 0 • 0 0000•e0 • • • • , • • • • • 0 , • •- . . 148 EXPERIMENTAL AREA AND BASIC MATERIALS AND • METHODS ..... ,,,....,...... ,,,.,.,...,..., 148 • (a) Experimental -procedures ...... ,.. 148
(b) Birds ...... 0 0 • •, 0 0 ...... , • • • • • • • 0 0 0 , 149 (c) Environmental variables .....,...... ,,... 150 (i) Temperature ...... 150 • (ii) Available natural food ..,....,..... 150
(iii) Rain 0 • • • • * . . • • • • • • • • • • . . . . . 150
(iv) Grass ...... • 0.0000000.,.., ...... 151 3. EXPERIMENT 2.1, 1G:9B POPULATIONS FOLLOWED BY • 90:1B AND OTHER 'GREEN-DOMINATED' POPULATIONS. 151
(a) Chronology ...... • • • 00 • •• • • • • • • 151 • (b) Materials and methods ....,..,..,..,.,,... 151
( c) Results ...... • • * • • • • , • 0 0 • • • • • • • • 0 0 153 (i) Summary of data ...... ,. 153 (ii)Apostatic selection ..,.,...... 154 (iii) Relaxation of brown preferences .... 156 (d) Discussion ...... •....,,.,.,..,...,..• 157
(e) Conclusion . . . S • • • • • , • • , ...... • • , . • . 159 APPENDIX B: TABLE 36 ....,., ...... 160 CHAPTER VIII THE RESPONSES OF INDIVIDUAL TURDIDAE:
PART 2 ...... * ...... • 0 0 • • • • 0 0 0 162
INTRODUCTION . . , 0 ...... • • • • • • , , • 0 0 • 0* .. 163 EXPERIMENT 2.2: 9G:lB POPULATIONS. EXPERIMENT 2.3 a b: 9G:1B POPULATIONS FOLLOWED BY 1G:9B POPULATIONS ...... 163
-11-
Page
Chronology . ...... • 4 • • • * • • • . . . . , 163 Materials and msthods...,...... , ...... 163 Results . , . . . . . , ...... ...... 165 (i) Summary of data •...... 165 (ii) Predation ..,..... *,••009005050090.00S • 166 (III) Imprint-pecking ....,...... ,. 170 (iv) Short-term preferences ...... 172 3. EXPERIMENT 2.3 (CONT.). PARTS c, dand a: TRIMORPHIC POPULATIONS 174 Chronology ...... •099000 175 Materials and methods ...... , 175 Results ,00.Oo00000.,o.O,,50.000.,,.D .... ,.... 175 (1) Summary of data .. . •... . • ...... * • 9 175 (ii) Predation ••,.•...... ,...,.,...... ,.,.. 177 4. DISCUSSION ...... * ...... * . . . . . 177 (a) Possible causes of colour preferences • .. .•. 178 (1) Relative conspicuousness ...... ,.,....., 179 (ii) Taste ..**.• ,...*...... 180 (iii) Innate preferences ...... 180 (iv) Permanent acquired preferences ...... 180 (v) Temporary acquired preferences ...... 181 (b) Searching images . ...... •. •...... 181 (c) Relevance of imprint-pecking ...... 183 (d) Diversity of behaviour and its implications,.. 185 (i) Diversity of behaviour ...... 185 (ii)Apostatic selection ,...... 187 (iii) Predators and area effects ,...... 190 (a) Introduction of third morph ...... ,...... 193 S. SUMMARY OF CONCLUSIONS FROM EXPERIMENTS IN CHAPTERSVII AND VIII ...... , ...... • . . 193 APPENDIX C: TABLES 42 AND 43 ...... 193
CHAPTER IX FURTHER EXPERIMENTS WITH GREENS, BROWNS AND INTERMEDIATECOLOURS ...... 197 1 . INTRODUCTION . . . , ...... 198 2 . PREY *500•*4*öO*OO..5.*.*..O.*.*D*O9..D*90..*O5*.O 200 3. EXPERIMENTS 1 AND 2 : TRAINING ...... ,.. 202 (a) Introductory remarks ...... ,,.,...... 202 (b) Plan , • ...... , . . . , ...... , . . , . . . . , . , . . . . . 202 (c) Materials and methods ...... 204 (1) Locations ...... ,.,.,...... ,,...... 204 Procedures . . • . . ...... . . . . . . 204 Predators ...... ,..,....,...,. 204 (d) Results ...... 206 4. EXPERIMENTS 3 AND 4: 'NORMALLY-DISTRIBUTED' - POPULATIONS ...... 208 Introductory remarks...... 208 Basic design of experiments ...... 211 (i) Prey 211 (ii) Presentation of populations ...... 211 (Iii) Changing the compositions of the populations 9SS***•* •*bO0••O• O.....,... . 212 (a) Practicability of methods ...... 213 - 12 -
Page
Preliminary study: experiment 3 ...... 215 (1) Materials and methods ,...... ,...... 215 (ii) Results and discussion ...... 215 Experiment 4 ...... . . . . 221 U) Materials and methods .. .. ,...... 221 (ii) Results and discussion ...... 223 5 . DISCUSSION ...... 228 Stimulus generalisation and the specificity of conditioning •.... .. .. •...... 228 Disruptive selection ...... ,.....,...,...... 249 (1) Maintenance of distinctness of morphs . 251 (ii) Maintenance of variability ...... 253 6. SUMMARY . . • ...... 6 • • . 254 APEENDIX D: TABLE 54 . , ...... 255
CHAPTER X GENERAL DISCUSSION AND CONCLUSION ...... , 258 1. WILD PASSERINES AS PREDATORS OF POLYMORPHIC SPECIES...... * ...... 259 2 • BAITS AS NATURAL PREY ...... • • • • • • . . . • . . . 261 30 EVIDENCE FOR APOSTATIC SELECTION ...... , 262 Conditioning to a specific prey •. • • • 262 Specificity of conditioning • ...... 262 (c)Apostatic selection ...... 4. CONDITIONS FOR APOSTATIC POLYMORPHISM .....,.., 266
Density . • ...... I • • 6 0 • ...... 268 Conspicuousness ...... • . . . . . 271
Palatability ...... a ...... 274 S. TYPES OF APOSTATIC POLYMORPHISM ..,...... 275
6 . CONCLUDING REMARKS ...... 0 0 • • ...... 280
REFERENCES ...... • • • • . • • • ...... • ...... 281
APPENDIX E: PUBLISHED PAPERS . .. 0..O666•6 ... 294 - 13 -
TABLES, TEXT-FIGURES, PLATES
1. TABLES Page qpter III: 1. List of birds in experiments ...... 52 to 2. Colours of green and brown baits ...... 52 Chapter IV: 3. Colours of red and yellow baits ...... 68 4. Grand totals of baits taken in
experiment 1 ...... 60 ...... 69 5, Numbers of baits taken by blackbirds during observed visits in experiment la 71 6. Numbers of baits taken by blackbirds and house sparrows during observed visits in experiments lb and ic ...... ,...... 73 7. Grand totals of green and brown small baits taken during experiment 2a ...... 75 8. Grand totals of yellow and red small baits taken during experiment 2b ...... Chapter V: 9. Preliminary experiment. Daily totals of it baits taken after familiarization ...... 86 10. Sites of experiments in series 1, 2, 3 .. 89 11. Dates of training and presentation of 1:1 populations in series 1, 2 9 3 ...... 90 12. Numbers of baits taken from 1:1 populations in series 1, 2, 3 by blackbirds and house sparrows •...... 91 It 13. Experiments 3,1, 3.2 : daily predation byblackbirds •0* • ... •...,.....•... 93 14. Numbers of baits taken by recognisable female blackbird in experiment 3.1 ..... 95 Chapter VI: 15. Experiment 1. Daily totals of baits taken . . , • , • ...... , , . . . , ...... 103 16. Experiment 1. Numbers of baits taken by blackbirds during individual visits..... 105 17. Experiment 2,1. Daily totals of baits taken ...... 107 tt 18. Experiment 4. Daily totals of baits taken 109 19. Summary of data from 9:1 experiments
(9G:1B + IG:9B) •.000.000•o ...... 114 20. Summary of data from 9:1 experiments (lG:9B -' 9G:1B) 115 21. Percentage browns taken on first days of predation on 9G:1B populations...*.*...* 120 Appendix A:22-32. Results of repeat 9:1 experiments ...... 123 to 22 Experiment 3.1 . . •o•oe.•.•0 •.oD.oeo .. . 124 it Experiment 3.4 ...... 126 of Experiment 5.2 ...... ,. 128 It Experiment 5.3 . . ...... 130 IT Experiment 5.4 ...... •...... 132
Experiment 6.2 . .. 0•O0O ...... ,. ..,.,... 134 Experiment 3.2 ...... 136
29 0 Experiment 3.3 ...... * ...... 138 Experiment 5.1 ...... • ...... * • • • .. . 140 Experiment 5.5 ...... 142 IT Experiment 6.1 .....,...... •...... 144 * 14 —
Page
Chqpter VII: 33. Experiment 2.1. Grand totals of baits taken and imprint-pecked by blackbirds A , B, and songthrushT •...... 152 it 34. Predation by B in experiment 2.1 ...... 154 ft 35a, 35b, 35c. Behaviour of A, B, T, when starting predation on greens ..,...... 155 Appendix 8: 36. Experiment 2.1. Full daily results for A, B , T ...... * ...... , ...... 161 Chapter VIII:37. Experiments 2,2a and 2.3a,b. Grand totals of baits taken and imprint-pecked by black- birds C, D, E, C, and songthrush T ...... 164 it 38. Behaviour of C in experiment 2.2a ...... 167 it 39. Proportionlaaten and imprint-pecked by C, D, E, C, T in 2.2 and 2.3a ...... 169 it 40. Sequences of like and unlike baits taken by C, D, E, C, T in 2.2 and 23b ...... 173 of 41. Experiments 2.3c,d,e. Trimorphic populations : grand totals of baits taken and imprint-pecked by blackbirds C, D, E, G, I, robins, dunnocks, house sparrows .... 176 Appendix C: 42. Experiment 2.2. Full daily results for C, D , E, C, T ...... •.o.... 0*e*t ...... ,,. 195 43. Experiment 2.3. Full daily results for C, D, E, C, I, robins, dunnocks, house Spx't'OWs • .....• a a a . a • • a a a a a a a o . a a a a a a 196 çpter IX: 44. Colours of nine types of green, brown and khaki baits ...... , , . * ...... 20]. Sites of specificity training experiments . 203 Dates of training and presentation of 1:1 populations ...... • . ...... . . . . 203 II Grand totals of baits taken from 1:1 populations after training ...... 205 48a,Specificity training experiments, series 1. Daily results for blackbirds and sinai]. birds ...... . . , , . . . a • , • • • . 209 48b.Specificity training experiments, series 2. Daily results for blackbirds and small birds . . . . . . . , ...... , ...... 210 49, Initial composition of 'normally- distrThuted' populations ...... ,...... 2]1 50'. Experiment 3. Composition of whole populations in each of 3 generations ...... 214 Predation by male and female blackbirds
in experiment 3 6*a • ...... ,...... 218 Experiment 4. Composition of whole populations in each of 20 generations ..... 222 Mean colour of baits taken by blackbirds E, J, K, L, M and songthrushes T's over every 3 generations in experiment 4 s...... 227
- 15 -
Page
Appendix V: 54. Number of baits taken over every 3 generations by E, J, 1<, L, M, and In experiment 4 ...... 256
2. TEXT-FIGURES
Chapter I: 1. Frequency of a morph in population plotted against frequency predated by • an'apostatic'predator 23 Chapter II: 2. Relationship between density of Acantholyda nemoralis and its proportion in tits' diet (from Clarke 1967a) 40 Predation by rudd on 2 and 3 varieties • of corixid bugs (from Clarke 1962a) 43 Predation by minnows on dimorphic population of artificial prey (from Smith
1967) ...... 0400.0 .....•• ...•...... 45 'Roneo'-copy of grid to show distribution of baits ...... 57 Chapter III: 6. Interrelations of Chapters IV to IX 62 Chapter IV: 7. Experiment 2a. Percentages of small greens offered plotted against percentages taken 76 it B. Experiment 2b. Percentages of small reds offered plotted against percentages taken 76 01 9. Percentage4yeliow seeds offered plotted against percentages taken (from Pough 1964) . * ...... , , ...... , . . . . . 79 Chapter VII: 10. Sites of plots in experiments 2.1, 2.2,
2.3 •.,e..o.o .... 0000.,.g,.0..e..o ... 00000 147 Chapter IX: 11. Experiments 1.1, 1.2, 2.1, 2.2. Percentages of 4's taken over every 5 visits from 1:1 populations after specificity training 207 It 12. Experiment 3. Morph frequencies in population before and after 3 generations
of selection 46000.e.,,..,....000.s..e,000 216 13. Site of plot in experiment 4 .., 220 01 14. Experiment 4. Morph frequencies in population after every fifth generation of selection . . . . . * ...... 224 re 15-21., show the numbers of baits taken by each bird in experiment 4 and the average composition of the population over: it 15. generations 0-2 ...... 231 it 16. generations 3-5 ••• .00.00000.0 ...... 233 It 17. generations 6-8 ... , ...... 235 it 18. generations 9-11 ...... 237 ti 19. generations 12-14 ...... ,...... 239
generations 15-17 ...... 0000 ...... 241 generations 18-19 .....,.....,..., 243 - 16 -
3. PLATES
Page
Chapter III: Standard brown and standard green
baits •Gt ... 0..O*0000tOG.000..OG.O 54 IV: Maximum density experiments. 1:1 populations of standard greens and standard browns ...... * . ...... 66 Maximum-density experiments. 1:1 populations of small greens small browns, and small reds, small
yellows ...... ...... 66 çpter IX: Appearances of the nine types of green-brown baits ...... 199 - 17 -
CHAPTER I GENERAL INTRODUCTION - 18 -
CHAPTER 1 GENERAL INTRODUCTION
1. COLOUR POLYMORPHISM AND ITS MAINTENANCE
Individuals of some species can be classified into a number of distinct forms living in the same population at the same time. Then the rarest 'morph' is present at a frequency too high to be accounted for by mutation, this variation is defined as 'polymorphism' (Ford
190). It usually has a genetic basis and may be detected at different positions a1onga biochemical pathway; from close to the action of the gene, as in enzyme polymorphism, to actually visible
alterations of morphological characters. At this end of the system, the colour or colour-patterning of the organism may be involved. Some colour polymorphisms (as perhaps in Biston betularia L., Kettlewell 1958) may only be transient, with one morph gradually
replacing another. Others appear to be relatively more stable. This latter, balanced, type poses the more interesting evolutionary problems, in particular with regard to how it is maintained. The possible mechanisms capable of maintaining genetic polymorphism
have been examined theoretically by a number of authors (e.g. Dempster 1955, Sheppard 1958, Williamson 1958, Maynard Smith 1962, Ford 1965). In practice, however, the relevant selective forces have rarely been
identified. One possible exception concerns polymorphic mimicry in
certain tropical Lepidoptera (e.g. see Clarke and Sheppard 1960 for
Papilio dardanus Brown, Ford 1953 for Hypolimnas misip L.). In these, some or all of the morphs mimic other more distasteful species. A mimetic morph will be favoured so long as its frequency relative
to the model remains low. Once the frequency of the morph increases
its advantage starts to diminish. A definite disadvantage will
occur when it is so common that its conspicuous coloration - 19 - becomes associated with palatability rather than the opposite. Other morphs may therefore be at a relative advantage and a balanced polymorphism can result. However, there are many examples of colour-polymorphism where mimicry is not clearly involved. They are to be found throughout the
Animal Kingdom. Because of lack of space it is impossible to give a complete or even representative, list and no comprehensive review exists. It is hoped that the following examples give some idea of the range of species involved:
Metridium senile (Coelenterata, Anthozoa), Fox and Pantin (1961).
Pomatoceros trigueter (Annelida, Polycheeta), F$yn and Cjden (1954). c!paea spp. (Mollusca, Gastropoda), various authors. phaeroma rugicauda (Crustacea, Isopoda), West (1964). Arinadillium nasuturn (Crustacea, Isopoda), Adamkewicz (1969).. Philaenus spp. (Arthropoda, Hemiptera), HalJka and Lallukka (1969). Homorocoryphus nitudulus (Arthropoda, Orthoptera), Owen (1965c). phiocemina nigra (Echinodermata, Ophiuroidea), Fontaine (1962). phphorus maculatus (Chordata, Pisces), Gordon and Gordon (1957). Eleutherodactylus spp. (Chordata, Amphibia), Coin (1950). Malaconotus spp. (Chordate, Ayes), Hall etal. (1966).
Colour-polymorphic species inhabit a diverse range of environments, both marine and terrestrial, and their distribution is world-wide. The degree of polymorphism varies from species with two morphs (e.g. the larvae of the moth, Geometra papilionaria L.) to species possessing almost infinite variation (e.g. the butterfly clams,
Donax spp.). Within species, there are often very distinct visible differences between the morphs. Indeed such variation has in the past created taxonomic confusion. Until recently, the different - 20 -
phases of the bush-shrikes (2a1aconotue) wore considered as separate species, simply because they looked so different (Hall at at. 1966).
Similarly, the 170 African species of the land-snail Limicolaria have now been reduced to 17 (Crowley and Pain 1970).
How, then, are those non-mimetic polymorphisms maintained? Faced with a lack of evidence for one mechanism or another, many workers have assumed that hetorozygous advantage is involved.
Documented examples of ouch views pertain to Copaaa app. (Cain and
Sheppard 1954a, Ford 196 1aeonotus app. (Hail ot'Ql. 1966) and the platyfish Xiphqphorus maculatus Gnthor (Gordon and Gordon 1957). It is true that there is definite evidonca that heterosia can maintain a polymorphism (for instance in the classic case of sick1ecoU anaemia, Allison 1936), but its ubiquity as a mechanism has yet to be demonstrated. With regard to colour' polymorphism, this lack of positive evidence may, admittedly, often be duo to the difficulties of distinguishing between homozygotes and heterozygotes in the wild. This is especially true for organisms, like Cepaq, where there is still ho satisfactory method of distinguishing the dominant genotypes.
Heterosis is by no means the only mechanism capable of maintaining a polymorphism. Other possibilitas have bean discussed in detail by Williarson (1958) and other authors previously mentioned (P-18)- Nora specifically, Clarke (1962a) has examined these mechanisms with regard to colour-variation, particularly in COpSQQ. He concluded that the colour and banding polymorphisms of these species ore probably maintained by selective predation, in the absence of mimicry. This thesis focuasa On such a mechanism. -21-
2. SELECTION BY PREDATORS
(a) General Before considering this possibility in detail, it is instructive
to examine some more gener'al aspects of the relationships between predators and polymorphic prey • Tha can best discuss these features with reference to the well-known land snails Copasa nomoralia L. and C. hortensis !1u11. These spa cioc are sympatric over a largo part of the British Isles and inhabit both grasslands and woodlands.
They are highly polymorphic for the colour and banding patterns of their shells, and the genetics of these characters have bén worked out (Cain, Xing and Sheppard 1960; Cook 1967; Cain, Sheppard and Xing 1968), In Britain, the main predators of Cepeos are the songthrush
Tuz'dus ph ilonielos Hart., various rodents, and rabbits • The most relevant to our problem are the former, since they hunt almost exclusively by sight, whilst small mammals are more dependent on olfactory cues. In addition, the thrush definitely possesses colour vision (van Eck 1939), whilst rodents and rabbits discriminate only in terms of black and white (UaUs 1942). Hence, mammalian predators selectively predate those morphs of C2gaea that contrast in tone with the background (Cain 1953) • It follows that the colour of a morph taken by a rodent is not necessarily the same as that attacked by a thrush under the same conditions.
Selection of palatable animals contrasting with their background is now well-accepted (see Cott 190) despite the assertions of earlier workers like Selous (1909) and Pearl (1911). Experiments on the moth, Biston betulaz'ia, are classics in proving that conspicuous palatable prey are picked out in preference to those harmonizing with - 22 -
the background (Kettlewell 1955) • Balanced polymorphic species are also subject to strong visual selection. The colours of -
C • nemoralis morphs often correspond with the coloration of the background, due to selection by thrushes (Sheppard 1951). A similar effect occurs in C. hortensie, though the correlation with the background may be due to a different genetic response by the
snail; for example, the commonest morph found on leaf litter in
beech-woods is not unbanded brown as in C. nemoralis, but five-banded yellow, with all the bands fused (Clarke 1960). Colour-polymorphisms
in other species are also subject to changes in morph-frequency,
probably due to visual selection (e.g. in water-snakes, liatrix app.,
Cam.tn and Ehrlich 1958; in Limicotax'ia martensiana, Owen 1965a). If polymorphism in Copaeia is maintained by non-visual effects,
then years of selection by predators should have produced the evolution of a uniform shell-colour appropriate to the colour and texture of the environment. Yet the morphs are still distinct in appearance, and have been for at least thousands of years.. Sub-fossil
Cagaea shells still reveal some of the discontinuity of the variation they possessed when alive, since the bands remain discernable (Diver
1929). Similar' evidence for the durability of colour-pattern polymorphisms has been found in other gastropods, for example, LiniCola±'ia (Oven 1966b) and 'felacuinantus (Ewers 1966). If sufficient specimens of true fossil polymorphic gastropods exist, the banding variation might still be visible. Distinct patterns can be retained on specimens as ancient as Devonian trilobites (Esker 1968) and even older Caphalopods and Bvzchiopode (Foez'ote 1930).
It is therefore tecpting to suggest that some typo of predator'- controlled mechanism is maintaining çpaea and other non-mimetic piymorphiams. t10 must now consider what form it is likely to take. - 23 -
-o a) I-
> U a) 3
Frequency in population
:iura 1 The fzt3qusncy of a aorph in a population plotted against the froqusney predated. The broken line shows the expected relationship if predation is random. A hypothetical ote!nplo of apostatic selection is given by the continuous line; the morph is over-predated at high frequencies and under-predated when rare • Note that this simple example doos not involve additional frequency-independent selection; the sigmold curve is symmetrical about the discontinuous line. - 2'. -
Cain at al. (1960) otatod that oco iorphe of Sovaca s oom to zsonb10 coipononto of their habitat and are therefore kept distinct
by predatoro. 0tan (1966a) haü presented a similar cao for a
polymorphic East Africanspecies - of graochoppor9 HcmorocoEMh uo nitidu1u. It was realised by Cain at al., hoover, that boloctian
on this baois could not m aintain a po1ytorphim, sinco it vould be
freu.ency-indopendont. In addition it has been argued that oco
morpbs (o.. bright pinks of çpaoa and pallids of I4tnicolearia
martonsiana) are con spicuous against all backgrounds (Ciero 196a, 1969; Otion 1960).
(b) ~goatnticsoloctionn
It is noi opportune to proeent a raothd of oolo ction by predato that may be capable of maintaining po1yiorphiero in the absence of aiaiczr. Theoretically, the system to oicplo. It rolico on the
protaico that predators preforontiaily ooarch for, and oat, the most conon morphs of a polymorphic s pecies. Under such conditions the
rerer fores, oven if relatively conspicuous, ohould be protoctad from predation and honco be maintained in the population (sea Fig. 1). The m.ch&nisn, to zeney*depndont since the fitnoso of tho onotypo (and therefore the gone) is invorcoly related to Ito frequency. That this type of selection can maintain a polyorphioin bee been demonstrated many times (Wright 1948 0 Li 1962, Heidco and Jayekctr 1963, Clarko and O'Doneld 1964). In this dissertation, the tom
apoetatio oolooUon' (Clarke 1982a) will be ueod for frequency. dependent selection that maintains a colourpattorn pólyophica, since ouch a uochanioci promotoc 9clpOOtaoy 9 . That is, it favouro those rorpho, or l apostatool that are vioually different froi the majority (Clarke 1950 1 .
'Thpoctasjo r0001965) is oynonyouo with 'apiostatic cóloction . - 'Apostatic polyaorphio' (Clarke 1962a) is pelyiorphioo eaintainodby apoetatic ooloction. -25-
Historically, the hypothesis was originally suggested by the
eminent naturalist Poulton (1884) to explain dimorphism in the
caterpillars of the Large Emerald Moth, Coometra 12apilicnaria. L.
The idea was first applied in relation to CeDaGa by Cain and Sheppard
(195a), whilst Maidens (1955) also suggested that birds have to learn that particular,2Maea visual patterns are signs of edibility and hence maintain the polymorphism. More recently, Clarke (1962a,
1962b, 1969) 6 has taken up the cause that apostatic ealoction is
maintaining the shell polymorphism of Caj2aoa, Since then, the hypothesis has been invoked to explain many other cases of colour-
pattern pdymorphisme (e.g. Limicolazia, Owen 1963a, 1968b, 1965a; a1aconotus, Owen 1967; Philaenus, Owen and -Yeigez't 1962, Helkka and Lailukka 1969; Monte, Safriel 1969; poUua, Schi$tz 1971). Payne (1967) has advanced the hypothesis that polymorphism in
cuckoos and raptors minimises the chances of recognition by host and prey species. If this were true, apostatic selection would be
acting 'in offence' Rarer morphs would be at an advantage over commoner forms because they would tend to parasitiso or destroy disproportionately more victims.
Moment (1962) pointed out that predators may be responsible for the nearly infinite variation found in, for instance, some brittlestar (Ophiuroidea) and Butterfly Clams (Donax app'.). In other words the population as a whole benefits because there is no repeatable visual cue for a predator to learn. Moment called this type of selection 'reflexive' because "the frequency of any one typo is dotenninod by a feedback relationship with all the other typos". Howevert it is difficult to speculate how ouch a polymorphism could evolve (apart from invoking group selection, Uynno-Edwards 1962), other than by frequency-dependent selection This fact seems to have - 26 - been realised by Li (1962) when reinforcing Moment's hypothesis with mathematical evidence. Colour-pattern polymorphisms, then, may bestow an advantage to rare morphs (Clarke 1962a) or to the population as a whole (Moment
1962). In more general terms, such discontinuous variation may be
classified as a type of 'protean behaviour' (Chance and Russell 1959, Humphries and Driver 1967, 1970) which proposes the existence of patterns of erratic behaviour whose function is to confuse predators. To the extent that the polymorphism is unsystematic, predators will have to learn that each morph represents food. The behavioural process of how they are believed to do this is known as the formation of a 'specific searching image'.
3. APOSTATIC SELECTION AND SEARCHING IMAGES
(a) Searching image Croze (1967, 1970) suggests that hunting by 'searching image', a term coined by a naturalist (von Uexkull 1934), is more or less synonymous with the experimental psychologists' 'choice from sample'
(Kohts 1923) and the human psychologists' 'matching by set' (Vernon 1952). These specialists have apparently named the same 'universal perceptual concept' • The common processes involved consist of three
successive stages (Croze 1967, 1970): (i) 'seeing' the sample, (ii) examining the choice situation, (iii) reaction to the matching
stimulus. These three components can best be illustrated with reference to ourselves, for searching images are continually used in our daily
lives. We become conditioned to looking for visual clues associated
with some kind of reward. For example, a blue jug on the meal-table - 27 -
may come to be connected with the idea that it contains milk, if in
past experience, it has always been filled with this liquid. This
psychological conditioning is equivalent to (1) above; seeing the
sample. It is only when a choice situation (ii) is presented that the previous acquisition of a searching image becomes apparent. This may arise when, say, two jugs of milk are presented on the table; one the blue, as before, the other a jug of a totally unfamiliar design. Provided that the reward cannot be directly seen in either
jug s the 'predator' will almost invariably take the familiar blue one; he is reacting to the matching stimulus (iii). The choice so made must be induced by the searching image per se, even if it is contrary to unlearned, innate preferences.
Von tJexkull (1934) was one of the first to recognise the possible existence of searching images, when he found that toads (Bufo bufo L.), fed on earthworms, subsequently snapped at matchsticks in a choice situation, and at ants or pieces of moss, after a diet of spiders. The speed with which a searching image can be formed, and its specificity, is exemplified by the work of de Ruiter (1952) on the effectiveness of camouflage in the stick-caterpillar, Ennomos al can era L. Caged hand -reared jays (Garrulus glandarius L.) and chaffinches (Fringi].la coelebs L.) were used as predators. When sticks were added to the cages, the birds initially took interest in these new components.of their environment and pecked at them.
After the birds had become habituated to the twigs, dead caterpillars were also added. For a long time these went unnoticed, but once one was found (often by a bird accidentally treading.on a caterpillar), others were also discovered. Initially, in,their hunt for larvae, caterpillar-like twigs were also pecked, but the birds quickly became more efficient. This was even true for those experiments where the sticks were from the specific food-plant of Ennomos. Thus a searching - 28 - image can be extremely specific. As de Ruiter points out, however, birds in the wild would not have such a limited hunting area, and would have more time to habituate to twigs. His results, therefore, do not necessarily refute the camouflage hypothesis. The literature contains other examples of work on searching images, mostly inspired by the painstaking ecological studies of
L. Tinbergen (1950). These will be discussed later in relation to the evidence for apostatic selection.
(b) Searching images and colour polymorphism Apart from the recent work of Arnold (unpublished), to be discussed later, virtually nothing is known about the behaviour of çp-hunting thrushes. Certainly there is no direct evidence that they hunt this snail by morph-specific searching images, let alone preferentially search for the most common morph, as is required by the theory of apostatic selection. The same argument can be applied to all other non-mimetic polymorphisms. Yet some supporting evidence for apostasy does exist. This will be examined in the next Chapter. - 29 -
CHAPTER II, THE EVIDENCE FOR APOSTATIC SELECTION - 30 -
CHAPTER II THE EVIDENCE FOR APOSTATIC SELECTION
1. INTRODUCTION
So far, attention has been concentrated on the problem of how
colour-pattern pàlymorphisms are maintained. It has been pointed
out that the visible distinctness of different morphs: is suggestive of selection by predators hunting by morphw.specific searching images
This characteristic carries little iYeight as evidence for apostatic
selection when considered n isolation. It is the purpose of this Chapter to assemble and discuss the more conclusive findings that support the theory.
There are two main lines of evidence. The first Is provided by considerations of the effect apostasis would have on two sympatric
species exhibiting near-identical, polymorphisms (Clarke 1962b)
The second approach is more direct and concerns observations of prsdator..prc3y relationships
2. MIXED COLONIES OF CEPAEA
A testable prediction can be made as to the effect of apoatatic selection on two Closely related polymorphic. proy. It is essential for this prediction to be valid that the two species should: a. possess
visibly similar polymorphisms, b. live. in the same habitat, and c. share the same predators. If they satisfy these criteria, it is quite liiy that predators would treat them both as the same Species. This would mean that effectively one polymorphism would be maintained by predators selecting apoatatically. - 31-
If a particular morph is at a high frequency in one species for a reason unrelated to its visible properties, for example - then, if apostatic selection is acting, the same morph in the second species would be at an advantage only when present at a low frequency. The polymorphism would be brought to equilibrium at a point determined by the relative visibilities of the morphs against the background. Uniform backgrounds may produce an
'apostatic equilibrium point' with a high proportion of one phenotype in both apcio. In general, thongh,'9a series of visually similar but ecologically diverse habitats should exhibit negative relations between the frequencies of equivalent forms in the two species" (Clarke 1962b). In the absence of apostasy, postive relations would be expected.
Strangely, the only species to have been investigated with regard to the above prediction are Cepasa nemor'alia and Cepaoa hortensia (Clarke 1962b, Carter 1967, Clarke 1969). A glance at the literature will rsveal that the existence of negative relations is a matter of soma controversy • Their presence eisa apparently revealed by a sequence of samples from 'open' habitats near Oxford
(Clarke 1962b). Hors there was a statistically significant negative correlation between the frequencies of yellow effectively unbandod shells in the two species. In woods there was also a negative correlation, though non-significant. Likewise, Carter (1967) has found similar negative relationships within habitat classes, but has questioned whether they are duo to apostatic selection • In addition he has criticised Clarke's original data on a number of grounds, but without providing an alternative explanation for the negative correlations • A reply to Carter and a summary of the evidence for - 32 -
apoatatic selection are to be found in Clarke's (1969) paper.
Space does not aUot a fuller discussion of the Clarke - Carter'
controversy, but it is evident that many more collections of mixed
çpasa colonies should be made before pronouncing the final verdict.
3. PREDATOR-PREY STUDIES
(a) Introducing the evidence
Until recently there has been a lack of direct research on
apostatic selection and consequently much of its earlier support
derives from other lines of enquiry. The evidence from Reighard
(1908), Pepham(1941, 1942) and L. Tinbergen (1960) are cases in
point. Primarily as a result of the last author, new impetus was given to research by. ethologiots (e.g. Cross 1967, 1970; Wi. Tinbergon
ot al. 1967) interested in the behavioural aspects of predation and by population geneticists (e.g. Clarke 1962a, b; Arnold, unpublished) concerned with the implications of such behaviour to polymorphic prey.
Despite the diversity of its origins, the evidence falls into two groups • The first concerns experiments whore predators feed solely on one particular prey before a choice situation • In the second, such 'training' is lacking. The distinction between these two designs is important, for' though the first may reveal a mechanism
for apostasy, whereby predators become accustomed to searching for one type of prey, the second compares more realistically with the natural situation. Here, predators have the chance of becoming
conditioned to the first prey-types they encounter. According to the hypothesis of apotatic selection, they should on average become trained to the commoner types. - 33 -
(b)
Associative learning is a ueU.idocuinonted form of behaviour
(see Thorpe 1956, Chap. IV). Predators, for example, rapidly learn to associate specific colour-patterns with distasteful prey and hence become conditioned to avoiding them. (See c .g. Blest 1957, Sexton 1959, Brewer et al. 1960, Duncan and Sheppard 1964). The evolutionary importance of this typo of behaviour cannot be disputed for it has led to the development of warning-coloration and mimicry in my unrelated species. Is the converse of this behaviour also true; can predators become conditioned to search in for Dalatable prey of a specific- colour pattern?
(i) General evidenc
The classic work of Luuk Tinbergen (1960) provides an ,affirmative answer to this question. He shoved, in a comprehensive ecological survey of titnlce (Parus spp.), that birds can become so used to searching for a particular prey that later-emerging species are overlooked for a considerable time.. All these prey were Insect larvae, mainly of the Orders tepidoptora and Hymenoptera, inhabiting pine trees. A variety of techniques, including direct observation of food brought by parent tits to their young, enabled Tinber'gen to analyse the yearly factors determining the degree of larval predation.
Although the density of the caterpillars was a principle component, their size, palatability and conspicuousnoss were also involved. By taking these into account he was able to study the GffectS of pr density alone. - 3I -
When, for oample larvae of Acantholyda nemoralis and Panolia app. first appeared in the tits' hunting environment, they constituted only a very small proportion of the food brought to the young. During the next few days their densities rapidly increased and than remained constant. Theoretically the tits should have quickly exploited such concentrations of larvae, particularly as these species were known to be highly palatable. Instead, there was a lag of aboui a week, on average, before the birds started to exploit the new food supply, whereupon the larvae suddenly appeared in the nestlings' diet. Tinbergon decided that these results could best be explained by assuming that the tits were hunting by searching image. When the larvae first appeared in the environment they were taken entirely by chance and hence infrequently; the tits wore still soarching for prey they were more familiar with. As the tits found nore Acantholyda and Panolie, they were rewarded and gradually built up searching images specific to these prey. Total acquisition of these specific searching images led to the rapid increase in predation of the novel larvae.
Similar evidence for 'ecological' searching images has been provided by Nook at al. (1960) and Cibb (1962) • All these studies contend that the formation of a searching image to a specific insect prey leads to an increase In its risk of predation.. The important point to note, however, is that until this time the prey enjoys a certain amount of protection. If such behaviour is widespread among predators, it should be repeatable under laboratory condilfsne. The earliest experimental evidence comes from Roighard (1908), working with the grey snapper Lutlanust griceus (L.) as the predator and dyed sardines Athorina laticops CL.) as pray. His aim was to toot ci now disproved hypothesis (see Lorenz 1962) for coral-fish coloration. - 35 -
One hundred and fifty eoa..caught snappers in e large outdoor
pool were fad on Shout eighty sardines dyed blue. They wore then given a choice of five blue and five unfamiliar (either normal, red,
green or yellow) sardines. During presentation of the two colour.
types Reighaxd noted the order in which the sardines were eaten • Shen
blue and normal fish were offered, the latter type was understandably preferred. The remaining three tests revealed an overall preference
for blue. Apparently the snappers had acquired searching images for blue-dyed sardines • Even in the face of a multitude of predators
inhabiting an open pool e the other colour types were at an advantage, albeit short-lived.
Unfortunately, Roighard 'a apparent evidence for searching images may be deceptive. As Craze '(1967) points out, blue was the only colour-type used for training the snappers. It is wrong to ignore the possibility that blue was a colour preferred as a result of learning unrelated to the training.
In a comparable experiment, Smith (1967) attempted to train individual minnows, Phoxinue phoninus L., on one type of artificial prey (either white or blaàk worm-M6 pieces of a mixture of lard and flour). One group of five fish ww fed solely on the dark type whilst five other fish were given light prey. For three successive days, each minnow was given a thirty minute 'feeding period' on twenty monomorphic prey, scattered randomly on the sandy floor (2 That * 1 foot) of an aquarium. Thirty-six hours after the lest training period, each fish was presented with a 'population' of ten light and ton dark prey distributed at random within the tank.
Subsequent observations of predation revealed that, on average, the minnows significantly favoured taking the familiar colour. - 36 -
Similar results have been obtained by Bukeija (1968) who has rigorously investigated the increases in risk of novel prey types when
introducod to nine individual sticklebacks Casteroateus aculeatue I.
in an experimental tank wdor controlled conditions • The finding most relevant to the present discussion is that when the fish were
accustomed to feeding on Tubifex worc&, larvae of Drosophila were overlooked in preference to the former.
The experiments described above load weight to the argument
that predators continue searching for familiar prey when a novel
typo is introduced into the area of search • Such conditioning has been taken to an entreme by Rabinowitch (1961 9 1968). In one
experiment (1968) 9 chicks of captive gulls LXUS argentatue and Larus dolawaransie, wore fed on one of three restricted diets • These were.* chopped earthworms, pink Cat-food and green cat-food. After
five days of training each individual bird was offered a choice between first, the familiar food and one of the others, and second, between the two unfa!njljar foods • In both species there was a very
significant preference for the familiar diet. when confronted solely with unfamiliar food, some of the chicks totally refused to oat.
Comparable results have been obtained by Capretta (1969) for chicko of Callus gaUus.
Rabinowitch's earlier (1961) experiments were even more dramatic. Sin domestic chicks were reared on 'silo' from hatching and eight were similarly brought up on whoat • After sin weeks the dieto were revereed • So conditioned were they that two of the silo-trained and five of the wheat-trained chicks refused to oat the unfamiliar food and actually starved to death. - 37 -
These results with young birds agree with L. Tinbergen's (1960) contention that a prey is not recognised as food until tts reinforcing value has been established through ozperienco. In this case, however, the avoidance of unfamiliar food was probably OxaWarated as a result of the predators being at a highly impressionable ago. A searching image should normally be more flexible, for it is to the predator's advantage if it cn quickly switch attention to a particular prey. The main stimulus involved in searching image behaviour appears to be the Colour (or tone) of the prey • Dawkins (1966) has shown
that domestic chicks will, on aquiring a searching Imagej hunt for prey on the basin of colour alone • Cross (1967) has demonstrated a similar effect in carrion crows, Corvus corona I. For example, when a white cockle covering a piece of meat was presented to a crow
('Abendogo') on a beach, it subsequently turned over two white extraneous shells, a piece of white fluff, a broken white mussel shell and a dead rabbit in addition to seven white cookies, of the experiment proper, and a black atone • The latter incident was presumably an example of a chance encounter with an unfamiliar object (see page 4-). In another study Craze (pcit.) enquired more deeply into the apocificity' of searching images. Crows were trained to Overturn mussel shells painted red and dintz'ithtod on a beach. In the choice situation, those were presented in equal numbers with another unfamiliar form. They were distinguishable from the standard shells by one of three criteria: colour, shape or surface structure. Generally, his results were that the chances of the unfamiliar Lore being taken depended on the degree of colour resemblance between it and the standard rod shell. Thus ro&yoliow mussels were overturned as often as the - 38 -
standard rods, tyhilot yol].o'a russe10 tere totally overlooked. In comparison with colour, shape and surface touture were loss Important visual cues; red cockles, for oxasplo, were treated on equal terms with rod mussels. Until now, we have discussed the evidence only in relation to birds and fish, animals that mainly hunt by the sense of eight. It has recently been demonstrated that olfactivo cues can also elicit aerching images in prodatorm • Soana (1970) has boon able to condition captive rodents (pdemus sylvaticus L. and Mus ausculu Ia.) to search for artificial prey with a specific scant. During choice situations the familiar type was taken in preference to prey that had a different scene but identical appearance • Soano points out that there is scant literature on animal scent polymorphisms (but see 11er 1878 for scant dimorphism in the males of the butterfly, Bttus polydamus). This situation may simply reflect lack of research. If future work reveals a wealth of animal species polymorphic for scent, then we must not overlook the possibility that such variation is being maintained by apoatetic selection. The main implication to be derived from all tha findings discussed above is that predators may be capable of becoming conditioned to specific variants of a natural polymorphic animal. That this occurs, is, as will be seen, far from conclusive.
(ii) Natural ôlymophic pre The only polymorphic species that has been investigated in this context is Cepaoa hortonsis (Clarke 1961)1.
Two pairs of hand-reared aongthruehea were housed in separate sixteen -foot square cages • Each cage was divided into a straw-lined
1Since the time of writing * the work of Don Boor (1971) has been published (Sec page 265 ). - 39 -
'collecting-area' and a 'thruh'.stone area'. Snails presented to the birds in the collecting area were taken to the thrush-stone site for breaking. Here the broken shells could be scored and counted. The experiment involved only two phenotypes, y0ilou unbanded (Y00000) and yellow banded (Y12345) morpho. Possible errors from differences in taste and behaviour were minimised. For three days the birds were trained on yellow unbendedin one cage (cage I) and yellow bandeds in the other (cage IX). They wore then presented with 'populations' of the two morphs in unequal proportions: 260 ?00000 's and 40 712 3L5 'e in cage I
99 and &O " 260 ' " cage XI.
On each day, the btokeu shells were replaced by the same uuthev of living snails with identical proportions of the two morphG. Hence the composition of the population was kept constan1. The experiment was terminated when about 700 snails had been eaten. It was then repeated, the only difference being that the straw was slightly darker than before.
Clarke found no clear evidence of conditioning, since banded forms were preferred in every case, oven when the birds had previously been trained on unbandeds. In only one instance was the preference statistically significant; this was in the first experiment, cage I. When the results for both experiments in cage II were added together, banded forms were again significantly favoured. A possible factor contributing to these results is that the thrushes may have become very efficient at searching for Cepaca (cf. do Ruiter 1952) since this activity was virtually continuous (Clarke, personal communication). For instance, the shells may have been detected on the basis of tactile cues. The experiments have never been repeated. - -
- 100 ACANTHOLYDA •--:
0 0 o 10 U.
li.. 0 •
I.. z ic hi U - • a hi CL
01 (Z.______- oo-o oe • • 001 0.1 10 DENSITY
The rltionship between the density of A=nth2l Zdg ooli8 md its proportion: in the tits tot. otQ that at low dnitioo tho lamm wero taken boo ofton then oupocted (continuous lino) cd at high donsitieo more ofton • From Claxko (1962o); hie Fig.7. - 41 -
(c) Do predators become conditioned to searching for the commonest types of a mixture of palatable prey?
We can conclude from the preceding discussion that predators may sometimes acquire searching images after several encounters with a particular type of palatable prey. If this prey is the commonest variety of a polymorphic species then selection will be apostatic. On the other hand, a rare morph may be the first type to be encountered by a naive predator. If the predator finds it unprofitable to hunt by using a searching image specific to this morph, it may possibly switch to searching for a commoner variety. Unless it does so, it will not select apostatically.
We must now examine the literature for evidence as to whether predatorsdo, on average, select in an apostatic manner.
(I) General evidence
Reference is again made to the work of Tinbergen (1960). Our interest centres on the situation when the tits were feeding on a number of larvae simultaneously. Tinbergen found that at low densities certain caterpillars were taken less often than expected. At moderate densities expectation was exceeded, whilst at very high densities tits prefer a varied to a monotonous diet and so dispense with searching images. Fig. 2 illustrates these findings, with regard to Acantholyda nemoralis.
Strictly, these data cannot be taken as evidence for apostasis, since Tinbergen was dealing with changes in density, not frequency (Croze 1967, 1970), - 42 -
For genuine apoetatic selection, as the density of Acantholyda. increased, so the density of alternative prey should have decreased proportionally. This in fact did not always happen. r. J.N.t1. Smith has drawn my attention to an additional criticioc of Tinborgen 'o work (including the experiments mentioned on p.33). Predators probably deliberately seek out areas of local high prey density (see, for example, Goso-Cuotard's 1970 data for feeding by the redshsk Trip totanuc3 L.) • If the tito had acquired searching images for areas of high prey density (as oppood to for opacific typos of prey), then Tinbergon 'o estimates for overall prey density may not have been representative of the areas visited by the birds. However, the work of Halting (1965) has allo demonstrated similar responses to increases in prey density. He used captive small maiinals, including deer mice. Peroocue leucopua Rafines quo, as predators and cocoons of the sawfly Neodriprion sertifer Geoff. as prey • 7hon alternative food was available, the nuober, of cocoons taken by the predators varied with density in a manner similar
to the response of tits to Aeanthol24 as shown in Fig. 2. More otperimontal evidonco comes from the work of Popham (19 141). He investigated the selective value of crypois in the corixid bug Arctocoriea (Sigara) distincta L • Adults of these hemipterans can be various tones of brown, depending on the shade of the
background on which the larvae develop. In one set of experiments two varieties were offered to tudd, Scardinius oz opbtha3ius L. One of the morphs matched the underlying sand, whilst the other was non-cryptic. In each experiment different proportions of the tx) varieties tiers presented to the fish and those ratios were kept 43 Pimutoo aa and
Cicio ü ramalyola o2 tho data fton ?hcn (191 and 1932). The Figwo end Locade ao takan ftvm C1azco 1962ci (Ttrj. Se ic O quivOIGUT to C3.ao'o F11. S md Tie. 2) comoopmda with hic Fits. 6).
100 • 90
30 V V / I70 '0 60 so
VV V • 40 V
- 30
• V 2
V • - 0 10 20 30 40 50 60 70 80 90 100
V PERCENT. OFFERED -
__3. The rosulto of pMdation by dd Scwdlniun. (LOUCIOCtn- on oidd bwjo jai otinet. n c' ooIoa oopozL t to c1 u..ty9oo of buo woro ood to the prodator in di3t proportlono. The bQek 4 Tiao og uali7oto colcur and too ('i' on the 001d Colo chat). The horiMatal MIG iivoe the popotion of an y eo1otypo ofovd, the vart ical cdo ohoo the tion Vkon by the r:dd. 0on ofreloc r000nt o1ouztyo 'i (czpptic) and csood ct1oo mpr000nt colour typo °1,° (akor md loco yptic). Each aoitnont, thIh vaDrasonto about 200 °podationo' Io thoveoo ropmeontod by tvo cic1o3, ono open ond one wwood. For ca anp1mation of the ctioio md dattod liziioo, oeo to1 • Date frozn Popba (11).
oo
90 / VV/
- 00. V' / V 70 Q
•• 60 0,
Q' I I_so •e • V V • • V Z 40 / V • • , I .
V - • 20 • V
• • V I , /4 00
- 102030405060708090 1 00
V PERCENT. OFFERED
Fiçuo 2. A diaaz ek11a to M g . 38. givInc the ra sulto 09 omhor oozieo ©g oxpipl=nta it th$.h cm additional ce1tour.tyo voo ictoducod
(001otyo InO, ovon dav~tev E4 1iQo cptio thm 11 1 ) .Thia ic reproscatod by the b1th cizcloo. Noto the ch maa g in aU colour-types,, froo advantao. oco at 10 onoioe to dLoc6vcmtn3oowneoo at bL onoo • Data fror payhm (1042). - 44 -
Constant by replacing oach eaten bug with another of the same colour.
Pophazn'o data have been z'eenalysod by Clarke (1962a). The.
results of tho first experiments are presented graphically in rig. 3a.
Clearly, the non-cryptic variety was over.prsdated at all frequencies. Using the data obtained from when the phases were presented in equal proportions, Clarke calculated the expected proportions taken from experiments involving unequal numbers of the varieties • These expected values are shown by the continuous lines in Fig. 3a. We can now see that the disadvantage of the non-cryptic morph decreased at low frouoncioo and incroasod at high. This effect was statistically significant. In a second series of experiments (Popham 1942) when another variety of corizid was added to the system, there was again a
definite advantage for a variety at a low frequency. Correspondingly, each typo became significantly disadvantageous when relatively more
abundant (see Fig. 3b). Hera appears to be clear evidence for apootasis. However the results wore obtained from only three predators, and Popham himself showed that cox'ixld bugs become restless when on a 1xatk -gmQ=,d of a colour differing from their own • These two factors may have imparted error to the experiments.
Smith (1967) carried out a sot of experiments along the sane lines as those of Popharu. In this case, minnows Phoxinus phoxinue L. were presented with populations of dark or light artificial prey.
He used large numbers of fish and offered different proportions of pray to each individual. Under such conditions, he was unable to duplicate Popham's results • This was apparently because the fish were so variable in behaviour; any apostatio component of selection - 45 -
100 0
0•
41 80-
LU A 0 0 0 60 oc 0, I-z , 0 40 , LU 0 A CL , , 20
0 , /
C, 60 80 100 0 A 20 40
PERCENT OFFERED
£ The reeulte of predation by minuous (Pho,dnuB hoxinus) on dimorphic populations of artificial prey • Each trial ie represented by two symbols an open symbol desiating the proportions of light prey offered and taken and a filled-in symbol denoting the percentage of dark prey offered and taken • The circles and triangles designate trials on two different backgrounds. pots the large variation in results compared with the expected valo (dotted line). From Smith (1967). - 46 -
would have been ccmplotely masked (see Fig.
Individual differences in predator behaviour have also boon shown
by Pough (1964. In an attempt to test apestatic oclection, be offered rod and yellow painted sunflower osode, in ratios of 9:1, to caged chickadees (Parus a. atricapiUus). However, the: heterogeneity in behaviour was not enough to conceal overall trends, for each bird tended to prefer the colour at the lower frequency. That is to say s the predators were behaving in a manner completely opposite to that predicted by epostatic selection. The prey were probably at too high a density.
This point will be discussed in more detail later (coo Chapter IV).
(U) Natural plvor1)hLc pr2
We are again forced to turn to Smaea the only opeciec that seems to have been satifactori1y investigated in the present context. Reference is made to some unpublished experiments carried out by Dr. R.W. Arnold (personal communication). He sot up artificial colonies
of .Cel2aea nernoralis at different sites in beech woods • These populatins each contained over 1000 snails and differed from one another with regard to the relative frequencies of the morphs that they contained. Only two varieties were prosentedt unbanded yellows and unbanded browns. Thrushes regularly visited the qopulations and their anvils were periodically visited in order to assess the predation. Yellows were always taken to excess • That is, relatively too many yellows were taken from populations where this colour was presented at low frequencies as well as from populations were yellows were cownon • But by taking this preference into account, Arnold was able to show that selection was in -47- fact frequency-dependent.
4 CONCLUSIONS
From the preceding review of the evidence, three main conclusions can be made with regard to:
conditioning to a specific prey; the specificity of conditioning;
apostatic selection.
conditioning to a specific prey Sight-dependent predators can, under certain conditions, become accustomed to searching for a prey of a particular colour or pattern. The frequency with which this occurs in nature is unknown; few of the experiments described concerned predators in their every-day environments. Of these studies, the evidence from Tirthergen (1960), Mook et al. (1960), Gibb (1962) are least reliable since selection may have been based on other cues besides the colour or pattern of the insect prey. The most satisfactory data are those from Croze (1967, 1970).
The specificity of conditioning Apart from the work of Croze (1967, 1970) little is known about the specificity of this conditioning. Yt this is of crucial importance, since the theory of apostatic selection assumes that predators acquire morph-specific searching images. Attempts at conditioning prddators to a particular morph of Caea have failed
(Clarke 1961), though the occurrence of such conditioning is implied from the results of Arnold (unpublished) and from Clarket -s (1962b) work on mixed colonies of pea species. - 48 - - -
(c) Apostatic selection
Even if eight-dependent predators do, in fact, become conditioned to searching for a particular variety of a polymorphic prey, there to no logic in assuming that they select apoetatically. Yet it is critical to the theory that, overall, they should hunt for the commoner vax'ietles.. Arnold (unpublished) has provided the only aatisfaotory demonstration that predators (individually or overall) select prey in a freque cy"depondent manner.
The not Chapter discusses the basic techniques that gore used in experiments designed to investigate the above and other related topics. - '49 -
CHAPTER XI! EERAL NATERIALS AND tZTHODS CHAPTER III GENERAL MATERIALS AND METHODS
o blackbird! sing me something well: Uhile all the neighbours shoot thee round, I keep smooth plots of fruitful ground thoro thou !aayst warble, eat and dwell.
(Tennyson 18e2).
1. INTRODUCTION
This research was specifically aimod at testing apoetatic selection,, and in this respect differs from most of the work discussed in the previous Chapter. Tho approach to the problem was basically simple: wild birds in their natural surroundings wore offered populations of polymorphic artificial prey.
2 • CHOICE OF EXPERIMENTAL SYSTEM
It must be remembered that oxporiontal conditions, especially in the case of animals in captivity, may distort responses to the: prey • In the present experiments the birds were in their usual habitats (gardens or fields) and were therefore sub]oct to environmental
Influences outwith the experiments per so. In this respect the method differs from those involving caged birds and is nearer the natural situation. Ideally, predators should be directly observed feeding on polymorphic prey in the field. Unfortunately, accurate observation and identification are difficult under such conditions. Predators are active creatures; a meadow-pipit feeding on PhLlaenus in long. grass, for example, would probably capture and swallow its prey almost instantaneously. Even assuming he saw this act an Observer would have great difficulty in iiemtifying the species of prey caught, lot alone its variety.
An improvement on this approach is to present predators with - 51 -
artificial populations of their natural polymorphic prey. In this way the site of the ouporimental population can be controlled and can be so designed as to be fully observable to the researcher. Problems of practicality ar's, hotievor, still involved. An important potential source of error concerns the possible relative behaviour (e.g. mobility) of different mor'pha • This night be ovmrcoma by killing the animals before presentation, but most tdld predator's do not normally eat their usual prey that are already dead. Katz (1937) tells of frogs that starved by the side of a heap of dead flies; they did not rocog Leo the motionless insects as edible. In zoos, penguins are usually fed on dead fish. During eighteen year's of observations on king penguins (ptenodytec atgonica) Gilloopio (1932) did not note a single case of them taking a fish from the ground. At feeding time over bird had to be fed singly by hand. It was decided, therefore, to employ artificial polymorphic prey. Those are immobile and 'novel' and have the added advantage that they can be made desirable to many species of birds • Moreover they overcome the problem of obtaining large numbers of spOcified phenotypes. (round-fce6ing passerines are ideal for prey-selection experiments. Apart from their abundance and ubiquity, their main assets are that they hunt almost exclusively by sight and possess colour vision (Puinphy 1948 9 Kettlete11 1955) and are extremely efficient at searching for prey (of. Hoppnox' 1965, for Turdus mirator'ius).
Many psssoriner3 undoubtedly food on polymorphic prey in the wild.
3. BIRDS
The species of birds involved are some of those normally to be
found feeding on lawns or fields. A full list if given in Table 3.
Further details of feeding habits and general behaviour can be found
In Uitbcry at al. (1938). (i Table :1
List of birds involved in experiments
[Ciwnon name c name
Blackbird Tur'dua mezula L.
Songthruab Turdus philontelos Hart (z T. ericatorum Turtan.)
I Robin Er.thacus rbecula Hat. Dunnock (hedge sparrow) Prune lie noda1arie Hirt.
House sparrow Passer . dconesticus L.
Starling Stua_u1aris. L. Blue tit !arus caez'ulous L.
Table 2, Colours of the green and brown baits based on the Nunsell co1oin System (1966)
Value Hue wona 7 Green I 7.3 GY 10
Brem I 5 YR
Note. The hue of a colour indicatso its relation to red, yellow, green, blue and purple; the value its relative lightness or gray value; and the chrosa, its strength. - 53 -
The cost important predator was the blackbird (for Ito hebits1 Gee Snow 1958). This was always tho ft, and cornetimos tho only, species to feed on the oporim3ntal populationo. In genera1 additional species participated in otperiisnt@ carried out on lama or in market gardens,, in urban end ouburben areas • On the other hands all the studies perforeod in rural fields involved blQClthirdi3 alone.
. PREY
In ncst experiments* the prey wore groan or brown cylindrical pollets t 0.7 an long X 0.7 an wide and a description of their appearance and manufacture now follows • Other typos of prey will be discussed in the relevant Chapters (IV and IX).
(a) NoMh_olo mt The marphe differed only in colour. Plate 1 to come oxtont
Illustrates this contrast, whilst Table 2 lists the difforencoc more- precisely by giving the scores of the two colours according to the 1mcoll Colour Sytot (1986). Green and brown wore chosen for the
original oxperlm3nts (Allen 1967,and see Chapters V and VI) because it was believed that they would bo representative of natural palatable
foods • The results of these eporioents dictated that the easo. colours should be employed in most of the ensuing work. 1th regard to other properties the morphs wore apparently identital. The baits were of a manageable else for both experimenter
and blrdei from my point of view they could be easily handled, and most passerines could swallow thorn whole • Care was taken to ensure
that the prey (or 'baits') were made of approximately equal size. - 34 -
Plate 1
2 standard brown and 2 standard green baits on a grass lawn • Each bait is 0.7 an (approx. a") in length. - 55 -
There was no evidence for large differences in taste (see Chapter IV).
(b.) Manufacture
The prey were made from plain flour ('Snowy Mountain') and lard ('White Cap') in a 5 to 2 ratio by weight (after Turner 1961).
Kneading produced a malleable dough of 'plasticine'-like consistency.
Artificial food dyes (varieties 'green' and 'brown', supplied by the Scottish Colour and Flavour Company) were then added, 11cc of dye per 1000 gins of dough giving the required colours.
Once the kneaded mass of dough had attained a uniform coloration, it was forced by hand through holes (0.7 cm diameter) drilled in a sheet of perspex. The long 'worms' so-formed were laid side-by-side and simultaneously chopped into 0.7 cm-long sections. Large numbers of baits were made at a time and were stored in plastic boxes at room temperature until required.
5. ENVIRONMENTS As already mentioned, the experiments took place in gardens or fields. The background was usually grass, though in some cases soil was used (see Chapters V and VI). To the human eye, browns appeared slightly more conspicuous against grass, while greens were slightly more conspicuous against soil (pp. 118-119). Rain dulled the colours and made the baits soft and mushy, causing them to disintegrate, so that they had to be frequently replaced. There was no evidence that wet baits were treated any differently from dry ones. Experiments at different sites had different environmental variables and these were often correlated with climatic conditions. The birds (with regard to both species and individuals) were not necessarily the same at any two or more given sites and, in general, predators may be more selective in summer than in winter (cf.Prop 1960). The backgrounc at different sites also differed; the well-hewn grass of a summer lawn - 56 - is a more uniform substrate than, say, a winter field.
Variation also occurred within experiments. There was no guarantee that the composition of predators was the same at any two points of time within a given experiment (e.g. see p. 96), unless the birds were colour-ringed (e.g. see Chapter VII). Fluctuations in background conditions were also liable to occur, particularly in summer, because of mowing, rain or drought. The last two variables affected the overall colour of grass, but the differences within each experiment were slight. Rain only prevented birds from feeding when it was very heavy. Snow lay on the ground during parts of some of the experiments described in Chapter VI, but the effect of this variable was remarkably slight (see pp. 118-119). In other words, the bait populations had many variables in common with natural prey populations. We shall return to some of them at later stages. Despite the environmental differences between experiments, the results were often strikingly similar.
6. TECHNIQUES
(a) Presentation of prey At some stage in each experiment a population of randomly distributed prey was presented to the birds. The baits were usually at a density of two per square metre (for the exceptions see Chapter
IV) and the procedure of establishing one such population was as follows: (i) Experimental ]?lot First, the area to contain the population was marked out. This comprised a rectangular or square grid of one-metre squares delineated by metal or wooden skewers • Each column and row was respectively numbere and lettered by upright skewered cards around the perimeter and, in Figure 5 'Roneo'-copy of grid used on 22.e.69 in ' L c 9 9 Io experiment 5.5, Chapter VI • showing the distribution of 180 greens and 20 browns. The squares with double lines represent itiinaiiiIulial the one-metre squares of the plot. Own -- Skewers were placed at points equivalent Uivaii IMOOMENIMMIMID to the intersections of the do1&.1nes. 3 The four quadrants of each metre-square are also shown • For details of the randomization of the spatial distribution, C see text. 5LORIMMOMPUM1001001501 1OnEyMMOUBMIUM1001001 E SNEED MMUNIMMUMEMI F MIM-100 On M! ii DD 1! 11111 - 58
some cases, along the middle lines • This facilitated identification
of each metro-square. Physical and practical. factors limited the size
of tho plot but generally it was 100 square metres, corresponding to £1 population also of 200 baits.
(ii) Randomization of py distribution
The randomness of the distribution of the prey was an important component of the experimental system and the methods by which it was attained deserve full discusio. The first stage was to make
'Ronoo'..copios of the grid (Fig.5) with oath metre-square divided into four quadrants. These copies were then used to record the distributions of the future populations. Randomization of a given population was carried out in two stages, as follows: First, the baits were randomized with respect to position. The four quadrants of each motro..square wore labelled clockwise from 1 to 4 and then from 5 to 8 • Thus each quadrant was represented by two numbers. Tables of random number's (Fisher and Yates 1963) were now employed. Different rows or columns of these tables were used for each distribution. As the random numbers were reed, the sequence of occurrence of the numbers 1 to 8 gave the spatial distribution of the baits. The following points were observed:
A • Tho copy of the grid was filled in row by row. B • The positions (with respect to quadrants) of two baits were marked per metro-square.
C • Theo two baits were never positioned within the same quadrant. This necessitated the rejection of random numbers that followed one designating the same quadrant.
D • Random numbers 0 and 9 wore ignored.
Once the future poitiénn of the baits had been established,, their - 59 -
distribution with respect of colour was computed. In the early
experiments random numbers were used. For instance when two colours wore to be presented in equal proportions (e.g. see Chapter V), half
the random numbers (0.-5) designated one colour, say green, and the
rest (6-9) represented browns • The colours of the baits were then
superimposed on their positions according to the Order of the random numbers that were road. However this procedure frequently resulted in
discrepancies from the required proportions, and the colours of some of the baits therefore had to be changed. This process required further randomization
In order to quicken the operation later experiments Involved the use of playing cards • Thus in 1 1:1' experiments, equal proportions of cards of red and black suits were oiaployod. One colour represented green and the other brown • Care was taken to ensure that the cards were thoroughly shuffled before use. They wore dealt one by one and the ensuing order of the colours was systematically superimposed on their positions shown in the plan.
During the preliminary experiments it wad discovered that the plots may not be searched uniformly by the birds; for instance, they may prefer one particular corner. These birds would be exposed to the morph-frequencies only within this particular area and such frequencies might differ from the prportions within the population as a whole.'
In order to combat this possible source of error, the grids of later experiments were divided into four equal areas • Each of these sections Contained identical proportions of the morphs. Thus in the case of populations with equal numbers of greens and browns, each of the four areas had baits in ratios of 1:1 • This procedure mm imi sod the possibility of large differences in morph frequencies between different parts of the population.
Using the plan as a guide, the prey were now spread within the - 60 - grid. To eliminate possible bias in the positioning of baits relative to ground cover, they were dropped into their appropriate quadrants from waist height. Further randomization was added to the system by frequent changing of the distribution. This was usually carried out daily and ensured that the birds were unable to learn the positions of the baits.
(b) Recording of data Each population was observed from suitable cover; for example, a room or a 'Land-Rover'. Binoculars were essential for watching the behaviour of the birds • When a bird entered the plot, the events were chronicled in the following order: (1) Time of entry. Species of bird involved and, if possible, its sex, the colour of its ring (see Chapters VII, VIII and IX), and other recognition marks, Positions of baits 'eaten' or 'imprint-pecked'. This was done by tracing the paths of each bird on to the plan and thus marking the positions' of the relevant prey. If all or part of a bait was eaten it was defined as 'eaten'. But if it was picked up and then dropped it was defined as 'imprint-pecked' (see Chapters VII and VIII) and 'uneaten';
The colour sequences of baits eaten and imprint-pecked
(actually recorded under (iii)).
Other points of interest, for example behavioural aspects relevant to feeding.
Time of departure. This was rarely possible to record accurately when more than one bird was feeding.
With practice, two or more birds could be observed simultaneously but when predation became too heavy only a few selected individuals could be followed. The overall proportion taken by the rest was - 61 -
deduced by subtracting the numbers of Observed eaten baits from the total numbers actually missing.
All experimental populations wore at some Stage observed in the above manner, but some (see Chapter VI) wore'also subjected to 'ueohservd' predation, This simply moens that they were left unattended for o few hours and then ozaminod on return, in order to determine the proportions taken.
7 • PLAN OF EXPERIMENTS
The campaign to test apôstatic selection evolved along two main pathways. The first involved determining whether birds can
boapTaS conditioned to search for a particular colour (see Chapter XI, pp.33 .39). Ujid passerines were fed on either greens or browns
end wore then presented with a mixture of both In equal proportions. After a proliminaty experiment, a sorioo of duplicates was carried
out. Later experiments were designed to test the specificity of
conditioning. Those last studies involved the presentation of additional colours of various shades of khaki.
The second approach was to discover whether untrained birds. preferentially search for the commoner colour in a mixture of two morphs (see Chapter II., pp.40-. 47). Populations of greens and browns, with one colour nine times as common as the other, were presented, These '9:1' experiments were repeated a number of times and zosulto were obtained for selection by both individual birds and groups of birds.
Finally, the two strategies converged on experiments involving the presentation of 'normally distributed' polymorphic populations containing greens, browns and various shades of khaki. Piptire 6. Interrelations of Chapters IV to
LOW DENSITY EXPERIMNTS
EXPTS WITH GREENS, PRELIMINARY EXPTS. DUPLICATE EXPTS. BROINS & INTEPJ'IEDIATE WITH GREENS & BROWNS. WITH GREENS & BROWNS COLOURS
1QjI1,1M1 DENSITY EXPERIMENTS . 1:1 popns. after ' 1:1 popns. after 1:1 popns. after training . training training on khakis
UCUS JOURS OP GREENS & BROWNS V V '-•' IALL' BAITS 4' Polymorphic 'normally distributed' popns. With & without previous training
9:1 popns0 Responses Introduction of a of individual birds 3rd (khaki) morph (Thrdidae) into the population
9:1 popns. Responses VII 1 VIII IV IV IV of birds en masse
( S.
S . VI 9:1 popns. Responses of birds on masse
VI -63-
All these populations tier presented at densities of tso baits per square metre • In an additional sot of ocperimonts populations were offered at much higher densities. One of the aims of this study was to test'unifying' selection (Pieloiski 19S9) which In GfEact, is the opposite of apostatic selection.
Figure 6 summarises the evolution of the work and rcalites the experiments to each of the following five. Chapters. It should assist the reader to visualiso the connexions between the several major typos of operimonts • Of these, we first describe and discuss those that wore carried out at high, densities.
0 - 64 -
CHAPTER TV WAXX11U.DENSITY EXPERIMENTS - 65 -
CHAPTER TV 1AXflU'DENSITY ERI1ENTS
1, INTRODUCTION
This Chapter describes experiments with dense populations of baits • Consider the case of birds feeding from a bowl containing a mass of standard greens and browns in equal proportions. If predation departs significantly from the expected 121 ratio then it can be assumed that for some reason one morph is relatively more attractive than the other. Let us nci imagine that the colour's are presented in unequal proportions with, for example, greens nine times as common as browns. By taking the frequency..indopondant colour preferences into account we can dorivo values for the expected predation. If aposteitic selection acts at such high densities then the birds should take relatively more greens then expected. However, the literature contains evidence suggesting that the
opposite is the case: rex's types are likely to be picked out from a group. For example, Piolowaki (1959 and 1961) has shown that
goshawks (Acobiter geniti1is L.) take white pigeons more often when most of the flock is black, and black ones more often when most
of the flock is white • Similar results have been obtained for other predator-'prey systems by Tiuborgen (1946), Willard (1966, quoted by salt(1967j. and Pough (1964). The experiments carried out by the
last worker wore supposedly designed to test apostatic selection and
will be discussed later in the Chapter.
The main aim, then, was to discover just how wild birds respond
to populations containing baits, at vary high densities and in unequal Proportions. This line of inquiry progressed from using standard
Croons and browns in the first series of experiments to employing -
Nitr Mir
jit -, 0-- 1~ 74
01 ter
.7,':. .. •,,.. 4h •'1'b, tg4 - • 4 1 ' ••' ; V 4L ,.
Plate 2
Maximum-density experiments (series 1). Populati on 1a 150 standard greens + 150 standard browns randomly distributed within a metal sieve 19 cm in diameter.
... . .., •:.-, • - 'L-' - : - .'-' r• •''
. :-..
.. •
• .•'--_ .. • b C d - • ,,,, - frI - ç . .L. -aVi .... ...... '-.•i,. fr' +' • ,i.if
.-.: ...r ••.- .
-- 1 • - -, •.- . ilk .
Plate 3
Maximum- density experiments (series 2). 1:1 populations: an lift, 70 small greens + 70 small browns; on right, 70 small reds + 70 small yellow.. The p.tri - dishes are 9 cm in diameter. - 67 -
Oma1.to' baits of thoco Cmd other Colaum, in tho occnd
2 •UATRXA1S MD MTHODS
(G) Sôzioo 1: The pOpn1LtiGe OZ ocandad amon id bn baits wom poonted
In ci'1a (19 cz deto) MotOl Slovesom WrZ Agemont that onood that ttoz did not collect. In late Uarch 1968 9 150 gmon and 150 barn baitø randomly diota'ib*tod within a eiovo (Plato 2), thL& wee then placed an the oda of a 3m outaido the Dpattent of ZooloMr. ui7e2r3ity of Edthu. The hattc woo voy &.eeoly pedtoft in fact
they wora at the amimm p000lble dcnoity for eocte lying en a plane
otfaco. Podia1137 9 thic poplaticn (la) tieD oeDinod and, after
MCOT&ng the ntm-baro of baits takes e, ww roccnotittod to Ito oriSinal cio. i?oqut cboervotlow frorn thQ a4jolaing building dotenined that the solo podatozo woro blacktArdo, probably jmt a malo and his tate,
After 6 dare of pxwentation o, nd CM intovai of about two toekc,
the 131 populaticatiea ropMcad by ttio othor3 of the oao oio one with gono nine tioo nowo awiinon than bmmc (ib) nd the other with bwnr nine tinco., M eomm ae zeono (lc) The eioo ccntining tbieeo populatione tezo placed one Lmtro apart and tholp poeitiono relativa to one mother wore FmqUmtly and rmd=ly changed. In othav reopoctci the operitont siee cied out co bofio. Dias the oihth day, the blbirdo uovo joined by an i otoinzblo ntmibov of hou000 rwe and the to of Lovedation r=o ocodin1y. The onper4cont leotod ovo' a
(b) Soioo2 The odent of ym&otlaa by h©optio during the 1ao oiiant - 68 - intoduod a eowce of aor. These birde tended to pock the bcito ithe -tham oat thom ho.o and small places tcto coqent1y loft In the 0iove3 • Th000 wom roughly meanotitutod into whole boite In oz to dorlvo en ootiiato of the nmzbara of whole baito eoton. In the ozinto that onood, thie pz'oblora was ocozo by pmcantlna baits of ouch a oio that botmeoparrms could otiailow them ho1o.. Theoo baits aara romd (app. 0.3 cm dito) polloto thich wom made foa otiafl piceaz of co1ouod 1exd and flour rou1dod botoen thwch and feiethor.. In oitparltmnt 2a they wom gmen and ban o boftro r, Thiot in 2b they wore yal1w end rod (oe Tb10 3 end Plate 8).
Tthlo 3. Scod by the HeMS011 cO1Ou SYOUM 1966 (oee p.52) the rods and yol1cio had values as chcrn.
Value Cbici
ST 10
Rod 7.$RP 10
The popuXotioe of theoo baits wore proentod in cicu1e' (9 cra in dierto) pleetic potidihoa with small ho3io drilled in thofr be!ee for drainage. Each dish hold 140 baits and the diohoo waro p'eacnted in a etaibt line and placed 50 cm apart.. During the oMorl=nte the populatieno ioo maintained ae before. Epoziment 2e was ot0tdd on 10th Nay 1968 k whom 1:9 0 9:1 and 1l ppu1atianc trara offood to the bfrda • Afto four days, 7:1 and i0 populations woz' oddod and the opoicant centinted for four mom deyo. EzpozLeat 2b an 20th tey 1960 and 2611cod the oazo plan ac Ito podece000. Table 4
Grand totals of green and brown baits taken from populations
presented during experiment 1.
Proportions offered Dates Predation (1968) G B
la. 1G:1B 31.3 - 604 202 226 Expected 1:1 214 214
lb. 9G:1B 22.4 - 27.5 1606 592 Expected 9:1 1978.2 219.8
Expected constant selection 2027.7 170.3
ic. 1G:9B 22.4 - 27.5 743 942 Expected 1:9 168.5 1516,5
Expected constant selection 215.9 1469.1 - 70 -
3. RESULTS
(a) Series ii (1) Synopsis
Table 4 gives the grand totals of baits taken from the 1:l and
two 9:1 populations of standard greens and brona. The values for expected predation, assuming random selection, are also shown. The
results for the 1:1 experiment suggest a slight preference for brown
but the deviation from expected is not significant Q 2,1.3S, P> 0.20) • On the other hand, both the 9:1 populations suffered
heavy overpredation of the rarer morph'. Those deviations from expected are very highly significant (when brown was rax'e, X1) 700.309 P ' 0.001; when green rare, X2 ) 2176.80 0 P c 0.001). Clearly, the birds were exorcising selection against the rarer colours and this was of such a magnitude that it overede any selection due to frequency-independent colour pro ferancos.
Nonetheless, another type of selection is in fact, detectable. The two populations containing the colours in w%oqual proportions
were carried out. simultaneously and should have been subjected to identical environmental influences, including numbers of visits by
predators. It is valid, therefore, to compare directly the data
collected from experiments lb and Ia. If this is done, we find that as well as preferring the rarer baits, the birds tended to ta}to green ones • This tendency can be detected by comparing the proportions of
common and rare baits taken frOm each population. The null hypothesis is that these ratios are the same, assuming constant selection against the rare forms, Thin is not the case (x 1) 128.51,, P CO.001). It can be deduced from the data that this heterogeneity is due to the
'Some of thoso results iavé now been pubifehod (Allen 1972). - 7]. -
Table 5
Nba9 of baits taken by ipale (O') &ad female () blackbirds during cbsorvd th3ite in oxperiinant la.
bsze5 taken Data observation
C B 31.3.68 19.01 0 3 09
14.00 12 0 0"
18.0 13 0 0"
3.14.68 11.10 3 0
11.29 2 0
11.31 7 0
4.14.68 9.30 0 17 a" 9.31 3 0
6.4.68 9.29 1
Grand totc-JD 39 27 predation of relatively too many greens from both populations.
We can show that this pr forence has, indeed, little effect on the general frequency-dependent findings described above,. If selection remained constant for the duration of lb and lo then the cross-product rating
no, of browns taken X no. of greens, offered no • of greens taken X no.. of browns offered should be the came in both experiments.. The average C.P.R. gives a measure of the relative preferences: C.P.R. = 0.756 9 which indicates that groans were the favoured prey. Taking this into account we can derive the expected numbers of baits taken. Reference to those values, shown in Table 14, reveals that selection against the rarer morphs is still excessively heavy (x 1 ) 1131.58 0 P 4 0.001 for experiment b and X 2 = 11475.87, P < 0.001 for experiment c).
(ii) Behaviour of the birds
The findings yielded by examination of the grand totals for each experiment are supported by the data collected during the periodic observations. Blackbirds accounted for all the observed predation in is and for most of it in lb and Jo. This is illustrated by Tables 5 and 6 which give the numbers of baits taken per visit during experiments is and Tht lc respectively. Altogether 1 the Observed birds romovd approximately equal proportions from the 1:1 population ends when feeding on the other populations, they tended to take the rare colours, despite an omnipresent preference for greens.
With regard to experiment la, if selection at each visit had been random, we would expect the baits to have been taken in approximately equal proportions. This was clearly not the case.
The results appear to be vary heterogeneous, though they are not - 73 - Tciblo 6
uibezo of baits takon by r10 (o') and Vomalo b1ckbids d bouo oporw=o (11$) dewing obsozvod vioit9 in epoztmonto lb find lo.
ext. lb oxyt , l 9:1 19 Data Tim of Numbem Numbozv obo'vition tabn takon (B.S.T.) G a C B
22.4.68 11.40 5 0 O of 14.43 1 0 23.4.60 16.24 .1 0
20.34 0 7 ? 24.4.38 11.46 2 0 26.4.68 14.45 2 0 O' 30.4.68 10.32 5 0 1.4.68 11.10 2 0 7.4.68 15.25 2 0 0"
15.28 3 0 Q' ' 7V ci IS .39 0 9.4.68 11.00 4 0. ? ci 16.10 2 0 US 10.4168 16.50 0 5 a" 12.4.68 17.30 3 0 cx" 91 2 0 ci 17.42 0 1 US 91 ii 1 OHS ci 1' 0 US cc 1 OH 13.4.60 12.00 1 0 0" 18.10 1 0 cP 14.8.60 10.00 3 0 cx" 15.4.68 11.01 4 0 c, ii 16.52 1 0 0 16 .4 .68 10.03 3 0 10.42 2 0 " 17.4.69 11.15 0 2 0 2 " 11.20 0 3 18.8.68 11.30 8 0 ci if 4 0 'I it 4 0 91 12.13 1 0 US 19.4.68 11.15 0 0 0' 20.4.63 14.43 1 0 US 23.4.68 10.02 9 0 0' " 10.03 3 0 'I 3. OH. ft it 1 OHS 26.8.68 ft 0 ii. O
Czand total 48 17 80 11 a Tho data for 11.30 B.S.?. 18.8.68 ropoort thzoo conoocutive visits by tho ao foto10 blackbird - 74 -
amenable to analysis by chiequered. Indeed, on only ttio of the nine visits did a bird take both types of prey. Thus after selecting
a first colour, a bird would tend to continue picking out the same type.
These differanesa Of behaviour were even more pronounced in the subocquant exporiaents (see Table 6). The results f*om forty..one
visits are shcrn. On average, at each visit 9 we would spact baits to have been taken In the ratio of 48 greens to 17 browns in experiment b and 40 greens to 11 browns in experiment c. In fact not one of the vicits involved predation of both colours. Again we can presume that
once a bird picks out a colour, it continues to select it Even when a bird visited both populations, as on 18.4.68, it still continued to Choose greens from both, although in one population this type was rare and in the other common.
(b) Series 2
The data obtained from these experiments support the main findings described above • Tablo9 7 and B summarise the numbers of baits taken
in experiments a and b respectively and Figs. 7 and 8 present these results graphically by comparing, for each population, the proportions taken tSith the proportions offorod.
The results will firstly be treated by assuming that selection from each population was random., With regard to experiment 2a, it
follows that too many of the rare colours were removed from each of
the four populations containing the mozhs in unequal proportions.. In three of them the effect was highly significant. Taking all four together, there was a significant tendency to take the uncommon colauro (Z •= 23.031, P<0.001). Predation on the 1:1 population - 75 -
Table 7
Cz3nd tota1 of green and bcn etail baits tekon fr'on
pou3.ationc poontod In onporlmnt U. Podato3 tore
Ppno. Pvodatlon Eectod pzdn P I ) 22ed (no 0100tion)
B B
IGIOD 212 362 57.3 516.6 62.66 cO..C31 SGOD 155 3.75 99.0 231.0 45.2S 0.0.001 1021B 260 202 271.0 271.0 0.89 NS 202 99 210.7 90.3 1620 US 901 1B 302 174 500.4 9.6 280.1 <0.001
Thbl 8
totalo of rodand yeUci oa3.1 baits taken from
op1ctiono p000nted in oeint 2b. Pmdators tsoo
b1ackbido and houoe opro,
P £poctod prodm. p Prcpn. Pxodation Ecpoctod pz'e X 1) of2ord (no oo1octio) - (an basis 09
V R V R V
LRs9Y 79 152 23.1 207.9 190.30 40.001 35.1 195.9 6.70 40.001 3Rs77 53.1 61 40 12 93.8 9.82 <0.02 5.8 79.2 0.097 NS R:V 142 08 115.0 115.0 12.60 <0.001 -o _ 7RaSY 84 50 99.8 42.6 7.99 <0.01 112.2 29.8 33.74 '0,001 Dfl1V 260 81 306.9 34.1 71.67 <0.001 319.0 22.0 3.69.40 40.001 - 76 -
was not significantly different from expected.
The birds responded to red and yellow baits in a similar fashion.
Again, in 2b, the four relevant populations all suffered e*coasivs predation of the rare colours and in each came this effect was significant. On the other hand s selection from the isi population
also differed significantly from xpocted.
Apparently, then, selection was f quencydepondmnt in both experiments, that is if we asso that the birds were expected to take the baits in the safle proportions as presented. Ue must now consider the effects of frequency-independent colour-preferences on the significance of those findings. Consider the results obtained
from, the it 1 populations of both experiments. Hero., frequency -
dependent effects should be absent and the values of the cr000produCt
ratios
bG b ) and (,) j( ry should represent the relative preferences for the two colours used in each study. !n 2a the values for b and g are not significantly
different and b 1.076, As can be predicted, when this figure is g used in estimating the expected numbers taken from each population,
the resulting curve scarcely difere from the straight line derived
from assuming random predation (see Fig. 7).
In experiment 2b, however, y a 0.62.
The expected numbers taken, assuming this preference to be constant
from population to population, are given in Table 7 and, as percentages,
In Fig. 8 • Three of the four relevant populations still show over
predation of the colours present at lower frequencies and these
deviations are sal highly significant. The overall tendency to take
Uhevo G. B, R, Y and g, b, r, y represent the numbers of greens, brcmo, rode and yellows offered and taken respectively. 77 -
Figure 7
gil / //
70 C a, ' A a
U) • C 50 // a, .// I- a)
30 /
10
10 30 50 70 90 /e greens offered
Figure 8
v7 ml 7/
-7- / 7/. 70 a,C /... a /_ 'n50 /./
• 30
ill] 7/ /
10 30 50 70 90 % reds offered
7 and 0 EpoLat 2G1 (cuoll gmow, oo11 bxom8) and 2h (orxU mdo Gu3liL yoUo) MOPOOtIV-03ri parcont0aw Of ono c1 mdo tco iLnot povwntagec offerid. The stajt (bzo!o) tLoo Ltacte the onpactod "Otionchipo mewdua rmd= predation. The cue rapmoon aWetod qmdctlon on tho b&aio o2 the data - 78 -
tho rarcr coloum 5o alec otill eificnt .3o278
/ P 4 0.001).
4. DISCUSSION
The ev.dcnee cbtthod ftora pdatc an 1:1 populationo bea not suøst cy largo differancas in the mlative ettetinoee of ene and bzowue (thcuh rod baits mq7 how) boon profew'od to yeUot). At other fauenoioo the birds, owrall, ton&Dd to take rolativoly
toe cony evocus. Thic izpliee that blockbido, the min pmdatom s favouod emenag a roi3ult that apparently contradiota infoxotLen galnod frorn other eWrimcats (o .g. these doearibod in Chapter VI) carried out with baite at lower denoitioc. B1abiz'e tedod to feed an only cue w1aur durina each vioit
(eeo OzpODiTontQ la, Th and Ic). ThotQb this ouCoetQ that they wera =Ing seGrdbina itoi3ee, tio cannot colotoly rule out the possibility that come or all of the W.rde were coloeting on the basis of proforoncoo otrineio to the oporitxint. Pough (196to l, in sorne e oineuto with Individual chic cdeoo, Pauo Etz'icziilus obtained similar rooulta • T3irde in ooparato cagee tsoro pzeoto4 with a aerico of thirty 1l choicee botoou rod and blvo or yo110 and white ooer seedne The pray wdro offered at danoitie of
100 per 10 inch .actuaro. He found oiificant difforoncee in the behaviour of the hirde • As in the present o oioanto each bird ten&d to pick out seoda of the eate colour es its original choice. - 79 -
Pedctic *a yoJt, cd Mid OMIM=Ov oadt to 1tiooip botoon to pocntoo o f yoliw soods taken by ci cedoor cmd tho porecintaMo oggbmd. Tho otroiGht (b'okon) tho is tho oaVcatod volatianship dctio. WhIlDt tho curm oaooto :axpetod porcontaSeG an tho bavio of date gom 11 opu1atioo. PQU3h 16e).
Ii1
7/ 70 a) ci tn 50 0. -e a) >
30
10 /
10 30 50 70 90 % yellows offered - 80 -
(b) Selection of ae.p The results Imply that wild birds, on avtrage, tend to select the colour offered in leti froquncy relatively mora often than the cor-men colour. This effect is opposite to that predicted by epoatatic selection, as is well Illustrated by comparing Figs. 7 and 8 with Figs.. 3a and 3b. However the theory of apoetatió. selection originally excluded very high densities-(Clarke 1962a).
For experiments specifically concerned with testing apoatatic selection, the reader is referred to Chapter VI • The maintenance of po1ymorphion by such frequency-dependent selection depends on the assumption that prodators hunting In comploc surroundings, encounter few numbers of prey at the same time. They should consequently acquire searching images that allow for greatest efficiency in exploiting the prey, and those prey-typos that deviate from this image will be protected. Theoretically, these will be the rarest ruorpha. In the present ezperimente the predators could see a large numbar of prey individuals at the same time. Comparable results have been obtained by other workers, notably
Pough (1964). He offered dimorphic populations of painted sunflower seeds to groups of captive chickadees. The size of each population was 100 and the seeds were .randomly distributed on 18-inch square trays. When groups of 6 birds were offered rod and yellow seeds in ratios of 9:1 9. 1:9 and 1:1, they favoured yellows at all frequencies, but also tended to take the rare coloure. Fig. 9 illustrates those results and can be compared directly with Figs. 7 and B. Hayes
(1962 0 quoted by Pough) obtained essentially similar results. Pough's original intention was to test for apootatic ooloction and ho explained his apparently anomalous results by suggesting that the densities had boon too high. - 81 -
The present work with wild birds and Pough'z, with chickadees,
both Support the general theory that when prey era dense predators
tend to osrt 'unifying' (Pie1aaki 1959, 1961) selection. In more
general toresunifying selection, where predators act to o1iainato variability in a prey population, may be regarded as an einp1e of
selection whose effect has been described as 'stabilizing' (Schta1hausen 1949), 'centripetal' (Simpson 1953) and 'normalizing' (Uedd.tngton 1957),
The behavioural cause of such an effect in a prey species is probably quite straightforward. In close groups of prey, any different or odd
types are likely to be conspicuous, whether this difference is due to coloration, location or behaviour. To the human eye, and therefore
presumably to the avian eye (Pumphrey 1961), the rare forms in my experiments appeared to stand out against the background of the rest.
It is reasonable to conclude that the birds tended to take these
types because they were relatively more conspicuous. This is in accordance with out knowledge that pro datora preferentially choose non-cryptic prey (see, e.g., Sumner 1934 and examples quoted by Cott 1940).
Such an overall trend was not the result of homogeneous responses by the birds • On each observed visit in experiments lb and lc,, for example, blackbirds removed either only greens or only browns and, as already mentioned, this behaviour probably depended on which colour was taken first • We can hypothesise that for a given visit this first-eaten bait was most likely to be of the type that was rare and there fore conspicuous. Consequently, on average, this colour was taken relatively more often than the other.
Unifying selection can explain the remarkable uniformity of animals that live in close groups, for example in flocks of birds and - 82 -
shoals of fish • The principle cm also be used to prdict that
polymorphic species should onbibit least variation ( in terms of the number of orphs) at voy higb densities. Polymorphic aniaslc do, indeed, coethiso occur in hiah numbers per =it cujoa ,, At Sanna Bay, Ardasnurchan, it in almost impossible to walk on the sand-dunes without t'oading on Capasa nsizoralis and the density, of haorotnci gcauda in parts of the salt-marshes of the Tyningheme estuary,
East Lothian, may be as great as 10 Isopods per square inch (r.D.J. Heath, personal communication).
Won (1963a, 1965a, 1985b) has studied the e1aticbetween density and the degree of polymorphism in populations of Linicolarje
r'tenaiana. He found that this snail was most polymorphic when it was commonest at dnsitioo of 100 per square metro • In abooluto terms, this figure is comparable to the densities used by Pough.
In effect, hoover, it is probably relatively lower. Consider the two environments. During the experiments, with sunflower seeds and baits, all the prey war's simultaneousiy exposed to the predator's
fIGH of view. They were distributed in a twoisdimensional, unobstructed plane. On the other hand, natural prey mno distributed in many dimensions within complex ouz'z'otxodings. A predator searching for
4mic01aria may, in fact, never oncornter many snails at the same time, oven if they are present in what to W3 scou like high denoitie. Indeed, such conditions may be more suited for apoatatic selection. WO will return (p.2684 to the possible relationships between selection and density, after examining the results of experiments carried out at lower densitio.
S. SUARY
In conclusion, the main results can be ourimarisod as foUotsa: - 83 -
Wild passerines en masse ,tend to select the rarer
varieties of closely packed prey.
This was probably an average result of heter,geneous
responses by individual birds,
The birds appeared to be hunting by searching image.
10 - 8' -
CHAPTER V CODITIOflflG TO GREENS OR BROWNS -85-
CHAPTER V CONDITIONING TO GREENS OR BROWNS
1. INTRODUCTION
The maintenance of colour-polymorphism by apostatic selection implies that predators can become conditioned to searching for one particular type of prey. In Chapter II it was pointed out that that wild predators there is little direct evidence to suggest behave in such a manner. This Chapter deals with attempts at conditioning wild birds to either green or brown baits.
2, PRELIMINARY EXPERIMENT
This early study has been described elsewhere (Allen 1967 9, Allen and Clarke 1968), but is included here for completeness.
(a) Procedures The work was carried out in a garden near Dalkeith, Midlothian
(O,S. grid reference: NT 352678). During six days in September, 1966, approximately 2,000 green baits were scattered on a background of
dark soil marked out in a rectangular grid of forty one-metre squares. A population of forty greens and forty browns was then presented for three consecutive days. Blackbirds were the only predators. After a lapse of one month, the familiarization was repeated
with brown baits, after which the birds were again exposed to a
population containing forty of each colour. During this period dunnooks, house sparrows and robins were feeding at the same time as
blackbirds. It was, therefore, only practicable to observe the latter. Predation by 'small birds' was deduced from examining the
baits at half-hour intervals. Table 9
Preliminary experiment. Daily totals of baits taken after familiarization with one colour alone: soil background.
Hours of Observed predation Date observation Blackbirds "Small birds" (BST) (8+) (many) B GB
Familiarized 6.966 14.30-21.00 50 (9) 9 (1) - -
with greens 7.9.66 11.15-19.00 111(13) 14 (0) -
8.9.66 6.20-18.22 153 (32) 75 - -
Totals 314 (53) 98 - Familiarized with browns 15.10.66 8.25-16.05 52 (8) 103 (20) 110 254
(From Table 3, Allen & Clarke 1968) - 87 -
(b) Results and.diooueaion
The results are given, in Table 9. After each training e3Qoio2 the blackbir& took more of the fii1ier. colour. The trend is highly oiificant despite hstorogonofty bettoen visits • This
point is well made by the figures in paz'ontheeeo in Table 9 9 which record the numbers of visits during which the specified colour uao
taken in etcooe • For each day the figures depart cuLdficently in the expected direction from a 1,1 ratio.
After familiarization with groans there was a significant increase in the percentage of browns taken by blackbirds over every five visits (P < 0.01 9 Spoazcan's rank correlation test), suggesting that the bLrds were gradually rocoieing the browns as acceptable food. During the single day after gaailiarisation with browns there was no corresponding increase in the nuaber of greens taken by blackbirds,, but the 'small birds' did show such a trend (P <0.01).
3. REPEAT EXPERX1ENTS
(a)
Subsequent to the first, three sories of cnálagous exporimento were Implemented. In Outline, they wore planned as followos
Series 1 (tYintor . 1967) comprised four experiments- at separate sites. At two of the locations, birds wore trained on browns and, at the other two, on greens, before being presented with a 1:1 population.
Series 2 (Easter 1969) was identical with series 1, except that it consisted of six experiments.
Series 3 (tYintor 1969) involved two experiments. In one, birds were fad on green baits before being given a 1U population (a). - 88 -
Next, they were trained on browns and again offered a 1 1 population (b). The second experiment was similar, except that browns were od in the first training session (a) and greens in the second (b).
(b) Materials and methoda (I) Locations All the experiments took place on grass fields near Edinburgh. Their sites (nearest farms or villages) and O.S. grid references are given in Table 10.
The skewer-marked eperimenta1 plots were erected near the perimeter of each field, where nearby hcdgoe or tress provided cover for potential predators. Each plot was a square consisting of twenty-five 3t3.-squaroo.
(ii) Procedures At each site, training was completed by scattering it with baits of one Colour, every day for a week. Approximately 2,000 were consumed at each site. On the day following the end of familiarization, populations containing fifty of each colour were presented. These populathne were randomly distributed and were maintained in the usual manner, by repeated replacement (see Chapter III). Table 11 gives the dates of the periods of familiarization and presentation of 1:1 populations. The oxpor'imonts of series 1 and 2 each contained one day's presentation of a 1:1 population, whilst those in series 3 involved offering Iti populations for three days after each training session. tYith one exception, observations were made from a car. from In experiment 12, the birds were watched in the open,! beneath a hedge. - 89 -
Table 10
Sites of opoiznta in oeicc 1, 2 d 3, giving nearest fagwo or villageo cud O.S. grid reftmncoo. All the eoineto wom c&ied out in fielde in Widlothlem. Scetlend.
8R0tN-TRtIWED GREEN-TRAINED Ept • Site Ept. Site Series 1 1.1 Ddon Nino, Roslin. 1.2 3oohiUs, Bwdiohou. TT 27363 NT 268672
1.3 Cpio1ac1 Ro3ooU • 1.1 DThouio Cheetezo, NT 302617 RosoeU. NT 303637
Soioo 2 2.1 Merton. 2.2 Ddon Neine, Rotin. NT 261702 NT 27642
2.3 B'ooiU, Buiehoe. 2.4 Hilleud. NT 260673 NT. 235667
2.5 Upper DcThouoio. 2.8 Thoto. NT 808635 NT 291610
Soiøo 3 3.1e Cpie1Aw, Rosete11. 3Jb (s 3.1a) NT 302617
3.2b (as 3.2&) 3.2a Lihoton. NT 272704 MMM
Tth10 11
Datco of traLyAng and pzooQnQtLon of 1:1 populations in coiinto Qf? oioo 1 2 and S. All dates am ino1Lv.
BROMmTRAINED .SRAXI1ED
Ept, Dcteü of Dtoo of Eapt. Diteo of Doü of training p36ontQt10 tiItnj3 pontatio of 110. of 11jo.
Sewjol3 1 1.1 12.11.67'. 19.11.67 1.2 18.11.67.. 20.11.67 18.11.67 19.11.67
1.3 90.11.67. 7.12,67 1.4 8.12.67' 16.124.67 6.12.67 15.12.67
So1oo 2 2.1 4.3.63'. 11.3.68 2.2 4.3.68'. 12.3.68 10.3.66 10.3.68
2.3 9.3.68.. 15.3.68 2.4 19.3.63'. 26.3.68
• 15.3.63 25..3.G8
2.5 18.3.68'. 25.3.68 2.6 24.3.68'. 31.3.63 24.3.68 30.3.66
Sioo 3 $.lc 20.11.69.. 28.11.69'. 3.1b 31.11.69- 7.12.69..
• 27.11.69 30.11.69 6.12.69 8.12.69
3.2b 13.12.69.. 19.12.69'. 3.1.0 3.12.69'. 10.12.69'.
18.12.69 21.12.69 • 9.12.69 12.12.69 - 91 -
Table 12 'S Humbew of baits taken from 111 populations by blackbirds
and house sparrows () after training, and estimated numbers
of birds involved (8hoin in parentheses).
BROWN-TRAINED GREEN-TRAINED
ozcpt. Nuere taken caxpt. Numbers taken B C B
Series 1 1.1 1 65 (2) 1.2 26 0 (2)
113 0 14 1.4 55 0 (2)
S3ieQ 2 2.1 0 111 (4) 2.2 92 0 (2)
2 2.3 0 51. (8) 2.4 86 (2) 0 31 (6)
2.5 0 58 2.6 41 0 (2)
Series 3 3.la 0 71 (2) 3.1b 129 81 (2)
3.2b 24 135 (8) 3.2a 106 0 (4)
Pwilminary 52 103 314 98 eWriment (80 (8+) - 92 -
(iii) Predators Apart from one experiment, blackbirds wra the only predators noted feeding. In experiment 2.3 house sparrows were also involved.
The'h numbers of baits taken by these birds were estimated as in, the preliminary eperimant. In each of the present studia, compared with the preliminary experiment fewer individual blackbirds were involved. This was probably a result of the differences in enironments; blackbirds are more abundant in gardens and woods than in open fields and hedgerows (Snow 1958). Estimates of the numbers of blackbirds participating in ech experiment are given in Table 12.
(c) Results Table 12 gives the numbers of greens and browns taken during the presentation, of 1:1 populations. The data for each experiment in the first two series represent the results from a single day's prosentation, whilst those of series 3 are grand totals of predation over three days. In addition to the results of eorios 1, 2 and
30 Table 12 gives the data from the preliminary experiment for easy refer-ace. In oil, fourteen populations were presented to birds after
separate training regimes. The effects of training are clear-cut- In each population s, the familiar colour has been predated to excess. That is to oay, during presentation of 1:1 populations, birds wore continuing to search for, and eat, those baits that wore the sacs colour as the ones on which they had been trained. - 93 -
Tthio 13
pOZDitO 3.1 md 3.2 . Doily po&i@
by b1bidO.
poIct 3.10 Epii.tot 3. lb
BRO1N-TAflED GREEtTRAXNED
Date Nunbors taken Data Humbera to (1969) G B (1969) G B
28 NOV O 24 7 Dc 53 9
29ov 0 33 8Dec 39 114
30 Nov 0 12 9 W 37 16
0 71 129 131
Epimmt 3.2b EzpezitQnt 3.20
BRO..TRAINED GREEN-TRAINED
Date Humberc taken Data Numbers taken (1969) G B (1969) G8
10 Dc 5 56 19 Dec 35 0
11 Doe 6 - 36 20 Dec 44 0
12 Dec 13 43 21 DeC 27 0
24 135 lOG 0 * 94 -
A further measure of the acquired preferences is given by the
fact that in ten populations the unfamiliar colour was totally neglected, whilst in the remainder only one and two were taken
(poziments 1.1 and 2.e respectively). What is more, the unfamiliar
colour was totally untouched in each of the three days of 1:1 presentation in experiments 3.1a and 3.2a, after the first training sosolons. In other words, in twalvo experiments there was a homogeneous response to the prey • It is legitimat% therefore, to compare otetioticaliy the observed and expected values of the proportions taken. The deviations from expected 1:1 selection are all highly statistically significant. Each has a probability of less than 0.001 that the result was due to chance alone.
The only experiments that revealed appreciable elimination of the uncestcmer'y varieties were 3.1b and 3.2b. Presentation of 1*1 populations after training in these studies had already been preceded by experiments 3.1a and 3.2a respectively;, The birds were therefore familiar with both colours by the time the second training period was implemented. Vet by the time of the subsequent presentation of 1:1 populations, they were apparently again conditioned to searching for one particular colour, but now the varieties involved were the ones that had been
'unfamiliar' in the first experiment. They continued to take an excess of these colours on each of the three days of presentation in each experiment (Table 13) • Nevertheless, in one experiment there was a significant incrociso in the proportions of the unfamiliar colour taken over ovejr three visits, suggesting that the birds were ro..leerning to search for these prey (r5 0.739 P 40.05 w for experimefl 3.11 0.69, not significant., for experiment 3.2b). In addition, there were significant differences in the proportions taken on the first and last days after the second training sessions, but not between any other
1:f - 95 - combinations of days 5,44147, P < 0.02 for experiment 3.1b; Xl) 5.057 P < 0.05 for experiment 3.2bL
The results for the two experiments of series 3 imply that preferences produced by 'a first session of familiarization can be changed by a second. Though it was difficult to identify reliably
unringed individuals, there was no evidence for changes in the compositions of the predators after any of the training periods. It is therefore reasonable to assume that individual birds underwent changes of preference. This argument is strengthened by observations
on a blackbird with two conspicuous white feathers in her tail. The numbers of baits taken by her in experiment 3.1 are shown in
Table 14.
Table 114
Numbers of baits taken from 1:1 population by a recognisable female blackbird in experiment 3.1
Numbers 'eaken G B
After brown-training 28th Nov. - 30th Nov. inclusive 0 24
After green-training 7th Dec. - 9th Dec. inclusive 43 - 96
;. DISCUSSIOI'1
('a) Comparison bat-Veen preliminaz and ur.eh6 att % eiients
All the above experiments apparently demonstrate the striking
effects of training. It will be noted, however,, that the results of the duplicate experiments are more explicit than those of the preliminary study. This small discrepancy may be a function of the
differences between the two typos of environments: fields and
hedgerows as opposed to gardens and woods • The 'former habitats,. support relatively fewer blackbLrds (Snow 1958) and the numbers involved
in the duplicate experiments were correspondingly lower. Probably as a consequence of their low densities, the compositions of blackbirds appeared to remain stable 'and confined to those with territories 1 in the immediate vicinities of the experiments. Thus each bird had a
high chance of feeding on baits offered during training.
Gardens and woods, in contrast, support a higher density of blackbirds, and throughout the pilot experiment it seemed subjectively
that new birds were periodically finding, the experimental food supply. Conaequently,some of the birds observed feeding on the l:lpulationa
may have eaten few, or none, of the baits offered during training. This could explain why some birds took unfamiliar baits, sometimes
oven to excess.
(b) Effect of natural Breferencoo .
So far it has been assumed that green and brown baits are normally equally acceptable to birds • Hence the expected selection
from the dimorphic populations was assumed to be IU • In fact,
1Tho relevance of territory and territorial behaviour are diacused in Chapter VIII, in relation to some 9:1 experiments.. - 97 - blackbirds Conerally soom to pz'sfog borno (ese Choptss VI VIII) end this natural predilection oo8d conoivsb1y account for the roaulta obtained after fmiiciition with this colour. Hwevor g zneo to enporimnto described in Chapter T uides this possibility tmlikely. Tho proscatation of dimorphic pop1ations •Lth bne nine tir.ee as eoon sa ens Savo rosulto for naive blackbirds (end OZio congthh) on six separate occasions (see Table 20) • The grand totals taken by these birds tevo 13 greens and 2002 bras a ratio of approulmataly WOO. In the brotmatrsintn oiaouts described above (neglecting the preliminary oXpori!snt
Md 3. 2be both of which invold an extra proiouo troining 1 grcn and 370 brot'ns ware taken, That is to say., the pportioue of brosno and greens taken we= almost twice as groat in the present 1*1 aWrimcrnta as in the specified 9l etwUes. Yet in the lattor, browns ware presented in ralativaly greater noro • It is therefore z'oasonthlo to conclude that brown training does in fact reinforce any natural predilections for this colour,
(0) Condition
Szarising the rMin results, wild birds continued to search or familiar colours after each of sixteen separate training sessions. These conclusions will now be discussed with regard to the findings of other workers and, briefly, in the content of results reported esowhore in this thesis.
The repeatability of thoo oxoricnta is lent further waight by sots similar work, using green and brown baits, done at Liverpool. by Sister S. Shutt (personal eoaunicstion) • She fed blackbirds (no other birds wore inlvsd) on green lard.iendflour pellets for a week, on a backgrotmd of rod tiles. Twenty browns and twenty groans wore - 98 - then presented at an unspecified density. Baits were not replaced when eaten and only a pair of birds were involved. On the first day, seventeen greens and only one brOtm were taken and on the next, only greens. On each of the succeeding two days the male blackbird was the first to visit the populations and took greens alone • Confronted with a scarcity of her preferred colour, the female was forced to commence eating browns. Similar results for wild birds have been obtained by Oates and others (quoted by Cook 1971, pp.87a88). The prey were red and yellow 'maggots! made of flour, fat, and colouring and they were presented in populations of 49, at a density of one per square foot. Predators were bousesparrows, blackbirds and starlings. After four days of training on one colour, both types were presented for four more days. Their proportions were either 90149 70% or SOi of the conditioning colour. In ten out of twelve trials, a relative excess of the conditioned colour was taken (see Cook 1971 9 Table 4.1). As in my data, there was a suggestion that the tendency to remove the familiar colour declines with time. It seems probable that the effect described in this Chapter is widespread in wild birds. Including myself, three authors (and probably others) have obtained similar results • Our data support
Tinbergon 's (1960) original contention that wild birds can build up searching images to prey and may consequently overlook new species,, oven if those are relatively common. Or',tf the two prey were morphs of a species, then we have a mechanism capable of taintaining the polymorphism. Further supporting evidence comes from experiments with captive birds. The work of Rabinowitch (1968) has already been described (p.3), but considered unsatisfectorg since ho used chicks of a very early age. Pough (1964), however, was able to demonstrate •0
the effects of conditioning on prey selection by captive chickadees. Birds were fed on either 100 yellow or 100 red sunflower seeds for two days before being presented with 70 of the familiar colour and
30 of the other. In other words, he offered 9:1 populationa overall as in his other experiments (see Chapter IV, pp. 78-80) except that the birds were obliged to encounter 100 of the common sort before presentation of the remainder of the population. In comparison with his earlier experiments, he found that selection was now closer to expected, assuming random predation. Presumably, the tendency to take rarer forms was being balanced by a conditioned preference for the commoner type. More recently, similar results have been obtained by Mrs. Pat Miller (personal communication) for quails feeding on red and blue pastry prey.
Other results comparable to mine have been found for fish
(Reighard 1908, Beukema 1968, Smith 1967) and mammals (Soane 1970).
Most of them (and also those with birds involved morphs that are visibly, very distinct. Green and brown baits are no exceptions to this rule. The morphs of many species are often similar in colour. Croze
(1967, 1970) is the only worker to have used visually similar artificial morphs and his findings will be discussed later, in conjunction with experiments of my own. The evidence presented in that section (Chapter IX)will lend more weight to the present findings. More support is derived from some of the findings described in the next Chapter, which deals with the responses of untrained birds when confronted with populations containing greens and browns in the proportions of nine to one or one to nine. - 100 -
S. SU'1ARY
In ccnc1uion, we can summarise the main results as follows:
In 8 experiments, idld birds tors trained to search for groan or bro'm baits.
t1hGn presented with a mixture of both In equal proportions they took the familiar colour to excess • This occurred in all experiments.
In 3 experiments it was possible to reverse the preference by carrying out a second period of training.
The results of the same 3 experiments indicated that the effect of conditioning gradually 4esoened with time. - 10]. -
CHAPTER VI 9:1 EXPERIMENTS: PREDATION BY GROUPS OF BIRDS - I o—
CHAPTER VI 9:1 EXPERIMENTS: PREDATION BY GROUPS OF BIRDS
1, INTRODUCTION
This section deals with attempts to determine whether 02H2sl of naive birds will proforontially search foz the Coummonor colours
In populations of groom and brown baits • The background to this line of enquiry has been discussed in Chapter II (pagss41 47). At first it was decided to present the baits in populations with one colour nine times as common as the other and later oporinments adhered to these proportions. Each experiment consisted of two parts. Birds wore first presented with populations having one of the colours nine times as cowmen as the other. After a few days they wore offered control populations with the other colour as the common variety. Such a design meat that two basic types of experiments wore carried outs they started with populations of either 9 greens: 1 brown, or 1 green: 9 browns.
2. PRELIMINARY EXPERIMENTS
Two of these studies were carried out as part of a B.Sc.
Honours project (Allan 1967). The results of the first experiment have since been further analysed and have been published (Allen and Clarke 1968).
(a) Exi,oriment 1: iresontation of 9G:1B poDuations followed by 1G9B popu]ations (I) Materials and methods
A lOm X lOm grid was laid on the lam outside the Department of
1This does not necessarily imply that the birds wore feeding in the same place at the same time. - 103 -
1c j9. 1. Da-Uy total of belttalcons
Table 1. DAILY TOTALS OF BAITS TAKEN: GRASS BACKGROUND Experiment 1 180 greens : 20 browns Hours of Observed predation Date observation Blackbirds Starlings Dun- H. Sparrows (BST) (6+) (2) flocks (2) (many) G B G B GB G B
30.7.66 9.03-1.0O 2 0 - - - - 1.8.66 10.30-18.15 61 9 0 1 1 0 - - 2.8.66 8.30-16.30 71 22 - 4 0 - - 3.8.66 9.20-15.15 71 17 - - 5 0 10 1 5.8.66 10.30-17.05 45 11 - - 14 0 86 2 6.8.86 5.09- 7.51 69 15 - - - 14 0 7.8.66 16.06-18.10 35 12 20 2 - - 57 0 Grand totals 354 86 20 3 24 0 167 3 Expected 9 : 1 396•0 440 207 23 216 24 153•0 170 Expected, con- stant selection 3490 910 152 78 234 06 154•4 15•6 Experiment 2 20 greens: 180 browns Hours of Observed predation observation Blackbirds Starlings Dun- H. sparrows (BST) (2+) (many) flocks (2) (many) Date 0 B 0 BOB 0 B- ___w 9.8.66 9.30-15.15 . - 0 7 0 1 15 42
10.8.66 10.30-17.30 0 5 - - - - 5 155
11.8.66 10.50-16.10 0 14 1 37 - - 14 183 14.8.66 15.05-18.10 0 13 0 45 0 2 4 136
15.8.66 11.40-14.10 0 13 0 44 - - 2 80
17.8.66 14.30-18.35 1 11 1 19 - - 9 74 18.8.66 10.25-17.40 0 24 1 36 .1 4 29 111 Grand totals 1 80 3 188 1 7 78 761 Expected 1 : 9 81 729 191 1689 08 72 839 7551 an.. 1jpvuvvu, .uI1 - stantselection* 3.7 77.3 45 1865 28 52 914 7476 * See text. The approximate numbers of birds involved In each experiment are given I at the heads of the columns.
Mood on Tb10 10 E11on and C1rco 198) - 1014 -
Zoology. Univorsity of Edinburgh. 1ithin the grid, 180 green and 20 brlctin baits wore distributed at random, two baits per metre square.
Observations were made from a first-floor windw and the population was maintained in the usual manner. During the experiment the grid es under continuous observation and a record was kept of tho numbers of baits token by the birds at each visit. After seven days of predation, the otperiment was repeated with a population of 20 greens and 180 brct4ns.
(Li) Results and jCcuacion The daily totals of baits eaten are shown In Table 15. Below the grand totals are given the expected numbers based on the assumption that there is no selection. The blackbirds and starlings took more browns than expected in both experiments. Dunnocks and sparrows, on the other hand, took more greens In the first experiment and more browns in the second. It is c1ear, therefore, that the birds did exercise visual selection and the direction of selection
could change within a few days. These changes did not seem to be caused by alterations in the colour of the background, which remained
apparently constant throughout the period of study.
It romaine to enquire whether the differences between the two experiments are related to the frequencies of the colours. On the basis of thGirand totals we can calculate for each species the expected numbers of prey eaton assuming that the selection remained constant throughout the period of study. If this were so the cross-
product. ratio:
ITo • of areons offered X 1o. of browns eaten Lo. of brownsofforod X No. of greens eaten should be the same in both parts of the experiment. - 105 -
TcMqj. oitnt 1. Numbora of bito tdcon by bla&-birft ding Ldivdu1 vio em 1uut 2 9 1966.
No. of baits taken Visit U. B 1 4 2 7 0 3 3• 0 4 3 0 5 9 0 6 5 0 7 5 0 8 2 0 9 o •io 10 24 0 ii 3 0 12 5 0 13 1 0 14 0 12 Total 71 22
(Pmm Tablo 2e Allan and Clca .968) - 106 -
The expected numbers based on assuming constant selection in both experiments, are given in the second row below the grand totals.
It can be seen that, in fact, all four species took a larger number of coznon forms than expected and a smaller number of rare ones.
Thus the birds appeared to select in a frequency-dependent manner.
Direct comparison of observed and expected numbers suggests that the effect is highly significant (()2 cs 20.19, P < 0.0001)) - This
comparison, however, is not strictly legitimate because of heterogeneity within the series, deriving from two aources Heteenoitr between days. For example, in the oocond part,
house sparrows took the two colours in different proportions on
different days(6) 59.15 0 P < 0.001). HeterogeneitY between visits • Even where there is no significant
heterogeneity between days, as with the rsulto for blackbirds, there
is strong evidence that the proportions differed from visit to visit. Table 16 shows the numbers of green and brown baits taken by blackbirds
on August 2nd, 1966 • Each visit by an individual bird is recorded
separately. The data are not amenable to analysis by chi-aquaz'ed, but it is clear that they are very significantly heterogeneous. If
the birds wore taking the baits at random, but encountering then in the proportions of seventy-one greens to twenty-two browns, than the
probability of a bird taking twelve browns in succession (as on visit
14) would be loss than 1 in 10. When this heterogeneity is taken into accounts the frequency"
dependent effects, although striking, cannot be regarded as formally
significant. - 107 -
Table 17 ELt 2.1. DaIU totqI6 of bafto ton
Pdatov two blackbirds one swgtbrmh
a. 20 gonoz 100 bron
Data Hour@ of obsavation 0boeod po&itio (1967) (G.1.T.)
18 1 1 7.48 16.30 1 58 19.1 7.50 16.51 o 49 20.1 7.40 a 17.00 o 38 21.1 8.25 17.00 0 54 22.1 7.35 17.00 0 38 23.1 8.20 17.00 o 55. G'aud totals 1 (1.1) (311.9) b • 180 gmaws 20 bno
Data Hours of observation Observed pdrthn (1967) (G.4.T.) C B
24.1 8.35 • 17.00 6 39 251 8.10 - 17.00 18 40 26.1 8.07 - 17.00 17 37 27.1 8.14 17.00 5 33 28.1 8.3,5-17.00 9 45 29.1 9.00 - 17.00 12 36 30 11 8.30 17.00 16 50 31.1 7.50 - 17.15 9 47 Grand totalo 92 327 (91.5) (327.5)
(Numbers in parentheses represent the grand totals expected on the assumption of constant selection). - 108 -
(b) Experiment 2.11: presentation of 1G-.9B populations followed by 99:1B populations
(I) t4aterials3 and methods
The predators involved in both parts Of this experiment were
two blackbirds and a thrush. All wore colour-ringed and results were obtained for predation by each one. Hero we shall consider
their combined predation, whilst their individual responses will be examined in Chapter VII.
The procedures are described fully in the next Chapter. umtharising.'thèm, on 16th January 1967 a 10n X lOIi grid was. laid down on a lawn at my Reading home • For six days the birds were presented with populations Containing 20 greens and 180 browns. They were then offered 180 greens and 20 browns for a further 8 days. Other populations were then presented but these do not concern us hero
(see p.151) • All the populations were continuously observed and they were all presented during most of the daylight hours (ace Table 17).
(ii) Results and discussion
Table 17 gives the pooled daily results for predation by the three Turdidao • Too many browns were clearly taken from the two types of population. If this preference is csswaed to be frequency- Independent and constant in both parts of the study we can derive the expected values shown in parentheses. The magnitudesof the deviations are clearly negligible.
The birds searched mainly for browns after the relative Composition of the baits was changed. These d.y results were not significantly
1Exporimezits 2.2 and 2.3 are described in Chapter VIII. Table 18 EQirnot 13, 22LU , totaU of ittkon
Date Hoa of Blackbirds Docko Robins • Bltm Tits } observation (2-4) (2) (2) (2) 0 BG. B 0 B C fl C , 20 oo: 2.12 12.15 - 16,20 0 21 - 1 2 - 9 180 bmune 3.12 9.15 - 16.30 0 26 - •- 1 4 - 23 4.12 9.15 - 16.45 o 35 0 9 - •- - - 16 80 5.12 9.25 - 16.80 0 29 0 2. 0 4 0 8 13 01 6.12 9.35 16.30 0 67 1 15 0 5 0 7 3 59 7.12 9.41 - 16.20 59 0 13 0 6 0 11 11 53 Girnd totaló o 232.1 39 2 21 0 26 47 296 Epect 1:9 23.2 208.8 3.5 2.3 20.7 2.6 23.4 33 300.7 Epectod, ci tct oo1otion 3.3 228.7 1.4 33.6 2.2 20.1 0.3 25.7 50,3 293.7 b, 100 amoum 9.12 9.35 - 16.25 16 50 1 2 3 1 3. 5 3.0 30 - 47 0 3 1 2 25 20 103 broio 10.12 9'.90 3.6.50 U 4 5 0 99 7 9 0 2 1 2 16 U (0 11.12 9.20 - 16.50 12 5 12.12 9.15 - 16.40 7 48 7 9 3. 3 2 0 29 Grand tot10 46 19*4 19 25 9 U 5 9 80 122 Zpeted, 1:1 120.0 120.0 22.0 22.0 7.5 7.5 7.0 7.0 101.0 101.0 Ezoctod. c.&t3nt selection 27.7 212.4 12.0 32.0 7.2 7.8 1.1 12.9 121.4 8016 e, 100 SmOaGs 18.12 9.40 -• 16.55 32 20 15 5 4 0 3 1 68 20 bzc 14.12 9.20 - 15.30 44 33 14 6 2 1 - - 90 3 OL 15.12 9.37 - 16.10 91 21 12 4 0 - •- 37 0 Grand totals 117 74 41 11 10 1 3 1 195 9 Epctod, 9:1 171.9 19.1 4648 5.2 9.9,.. 0.1 3.6 0.9 130.6 15.4 Epocto, ecnott aelectim 103.1 87.9 90.2 11.8 9.8 ^ 1.2 1.7 2.3 193,4 10.6 - 110 - heterogeneous (X 5) .3O, P > 0.50) despite variation of the individual responses. Tuo birds searched for browns almost exclusively when rare, but the other relaxed its preference On the first day after changeover. We shall return to these birds in the next Chapter.
(c) !_er.tiaent H presentation of IC:9B pop1ations followedbi 1glB and 9G, ID. populations
This expoimant, carried out in the late winter of 1968 9 was essentially a repeat of experiment 2.1, except that 1:1 populations were offered between the ii 9 and 9:1 populations e-
(1) materials and methods
On 2nd December 1968 a grid of 100 metre quarea was laid down on the lawn of a garden in Dalkeith. For seven days birds were presented with 20 greens and 180 browns. After a lapse Of one day, equal numbers of greens and browns were offered for four days., followed by three days of 180 greens and 20 browns.
The baits when presented were observed continuously; blackbirds and house sparrows were the main predators but robins, dwmnocks and blue tits were also involved in all three parts of the study. In addition, later populations. were visited by great tits (Parue major I.), coal tit (P. atop L.) and starlings.
Results and discussion Table 18 shows the daily totals of baits taken The only results that reveal significant heterogeneity in the proportions taken per day are those for house sparrows in pert b. 3) 16.93,
P <0.001). As in experiment 1, however, the blackbirds exhibited considerable variation in the proportions taken per visit when feeding on populations presented during b a and c. - ill -
Below the grand totals are given the expected numbers based on random predation. The birds were clearly discriminating between the baits; thus the blackbirds, dmnocks, robins and blue tits took more brans than expected in all three parts. The house sparrows took too many greens in a. and c. and too many browns in b. The Table also gives the expected predation baod on the assution that selection was constant in a • and c • Those values wore dst,od in the same manner as before C see page 1044. In all ton cases the commoner colour was taken more often than expected. Overall this affect is statistically significant 5.7 9 P C 0 102). (( n The data from the 1:1 populations need some covmont • In every case browns were taken to excess on the basis of ll ratios, suggesting that the birds carriod'-ovot' preferences loareod from part (a) of the experiment. Overall this effect is statistically highly significant ((W2 53.07 9 P t o.00i) . On the basis of constant selection in parts (a) and (b) we find that robins and house sparrows took more browns than expected (X 1) 2.8030 P ) 0.05 0 not significant; X2 33.399 P < 0.001) • On the Other hand the 1) blackbirds and dunnocke for some reason took more greens than oXp6ated (X 13.76, p